Generative Artificial Intelligence in Secondary STEM Education in the Light of Human Flourishing: A Scoping Literature Review

Generative Artificial Intelligence in Secondary STEM Education in the Light of Human Flourishing: A Scoping Literature Review | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Generative Artificial Intelligence in Secondary STEM Education in the Light of Human Flourishing: A Scoping Literature Review Alissa Fock, Hans-Stefan Siller This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6923010/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 01 Dec, 2025 Read the published version in International Journal of STEM Education → Version 1 posted You are reading this latest preprint version Abstract Background: Education for Human Flourishing aims to empower students to develop their full potential to lead a meaningful, autonomous life to the benefit of themselves and society at large. Recent technological developments call for an evaluation of the entanglement of (education for) Human Flourishing and Artificial Intelligence. Following the PRISMA guidelines, this scoping review investigates to what extent the current research on generative AI in secondary STEM education provides a solid basis for exploring the interconnection of Artificial Intelligence and Human Flourishing in STEM education. To this end, 183 eligible publications were analyzed regarding their general characteristics, research themes as well as strengths, weaknesses, opportunities and threats (SWOTs). Results: The scoping literature review reveals a focus on cognitive aspects of STEM education despite the need to broaden human capabilities in the light of generative AI. Ethical aspects are sidelined, although the SWOT analysis shows some significance of these issues. Moreover, there is a lack of research on STEM-specific theoretical frameworks and research is concentrated in the Global North, both of which might undermine an unbiased, culturally diverse development of new solutions for generative AI in secondary STEM education. The majority of current research examines AI-generated content instead of human participants, and publications focus on the performance and development of AI tools instead of their impact and application. This might hinder a human-centered approach to AI in secondary STEM education, potentially threatening human identity and meaning and thereby Human Flourishing. Conclusions: Based on the results, we show that existing literature does not yet provide a suitable foundation for Human Flourishing related to Artificial Intelligence in secondary STEM education. Our findings thus point to future research perspectives necessary to strengthen Human Flourishing in STEM education and ensure a human-centered, meaningful approach to Artificial Intelligence. Artificial Intelligence Human Flourishing SWOT analysis scoping literature review secondary education Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 1. Introduction The adoption of emerging technologies in STEM education has long been double-edged. With each new tool – be it the pocket calculator or the rise of computer algebra systems – comes the debate of its promises and perils in the classroom. Regarding pocket calculators, which eased calculations in various STEM disciplines, concerns have been voiced already in the 1980s that these tools “will discourage students from using their minds” and that school will “graduate people who are unable to think quantitatively and who depend on calculating devices” (Willoughby, 1985 , p. 90). These fears of overdependency and loss of fundamental skills stood opposed to great expectations of calculators allowing students to tackle more complex and realistic problems without being constrained by long computations (Pendleton, 1975 ). With advancements in Artificial Intelligence (AI) and its potential integration into STEM education, arguments and debates surrounding earlier technologies appear to resurface. The term “Artificial Intelligence” is notoriously hard to define (see, e.g., Russell & Norvig, 2021 ). A subset of AI, which has seen a massive surge in public attention since the release of ChatGPT in late 2022, is Generative AI. Generative AI refers to AI systems that are able to create new content (such as text, images or videos) based on their training data (Taulli, 2023 ). Following the current discourse on generative AI – particularly ChatGPT – in the classroom, we notice striking similarities between statements made today and those on former educational technologies. Still, educational institutions perceive that the emergence of generative AI “ feels different” (Korseberg & Elken, 2025 , p. 960). Developments regarding generative AI fundamentally shake the role of human beings in the teaching and learning process. Probably more than previous educational technologies they impact human decision-making, identity and the purpose of being human – in other words, they affect Human Flourishing , the ability of humans to lead a meaningful, autonomous life by developing their full potential. To ensure that the adoption of generative AI in education does not impede but foster the flourishing of students and teachers alike, its effective, ethical and evidence-based implementation is of pivotal importance – throughout different disciplines and educational levels. This scoping literature review provides an overview of the current research on generative AI in secondary STEM education 1 , focusing particularly on research themes addressed in the literature and on the strengths, weaknesses, opportunities and threats of generative AI in secondary STEM education emerging from empirical evidence. Based on the findings, it is investigated to what extent the current research provides a foundation for exploring Human Flourishing and AI in secondary STEM education. 2. Theoretical Background 2.1 The Notion of Human Flourishing in Education Human Flourishing has been identified by various sources as one central goal of education (Curren et al., 2024 ; OECD, 2023 , 2024b ; Schinkel et al., 2023 ). The exact definition of the term depends on philosophical, psychological, political, academic and cultural backgrounds (Curren et al., 2024 ; de Ruyter et al., 2022 ). Broadly, the term Human Flourishing refers to the ongoing realization of a person’s fullest potential and living well as a human, which includes engagement in meaningful relationships and activities (de Ruyter et al., 2022 ; Wolbert et al., 2015 ). A person leading a flourishing life is able to act and decide freely and independently (a capacity referred to as agency ) (de Ruyter et al., 2022 ; Stevenson, 2022 ) and “has found meaning and purpose through autonomous decision-making” (OECD, 2023 , p. 22). Flourishing people contribute not only to their own flourishing but to the flourishing of other people in society (Curren et al., 2024 ). Non-Western accounts of human flourishing particularly emphasize this notion that individual flourishing is deeply relational and fundamentally connected to the flourishing of one’s community (OECD, 2023 ). The aim of education for Human Flourishing is to “increase [students’] ability to choose and follow their own path towards a flourishing life” and actively engage in creating and upholding circumstances that allow flourishing for everyone (Ergras et al., 2022 ), considering both students’ present and future lives (Schinkel et al., 2023 ). This requires a holistic view of the student, extending beyond cognitive capabilities to socio-emotional and personal capacities, targeting students’ critical reflection, ethical decision making and perspective-taking (Flook et al., 2022 ). Human Flourishing consists of intertwining scientific and moral reasoning, fostering both adaptive problem solving – including creative and critical thinking, synthesis skills and collaboration – and ethical decision making, i.e., balancing the needs of various individuals and making fair, altruistic choices (Stevenson, 2022 ). Successful education considers that students differ in their potential, wishes and thoughts about a desirable life, and recognizes different, individual ways of flourishing. While education should nurture these capabilities to enable Human Flourishing, flourishing students and teachers also enhance the quality of education (de Ruyter et al., 2022 ). This implies that not only students’ but also teachers’ flourishing needs to be considered (Curren et al., 2024 ). 2.2 Human Flourishing and AI in Education As students’ present and future lives are and will increasingly be infused by technology, technology plays a crucial role in students’ flourishing, enabling them to be responsible citizens in a digitized society (Ergras et al., 2022 ). While this holds true for a variety of technologies, the recent spark of attention in AI calls for a more detailed view of this particular technological development and its implications for Human Flourishing. To this end, the OECD ( 2024b ) has put forward three themes: AI and broadening human capabilities , AI and developing new models for the future , and AI and restoring meaning to individual lives . We take these three themes as a starting point for elaborating on the interaction between Human Flourishing and AI in education. 2.2.1 Human Flourishing Education and AI: Broadening Human Capabilities State-of-the-art (generative) AI systems are capable of tackling tasks that were previously reserved for humans, such as producing natural language and solving mathematical problems. These recent developments of AI underline the importance of fostering students’ socio-cognitive and meta-cognitive rather than purely cognitive skills, cultivating a holistic view that is distinct for human intelligence (OECD, 2024b ; Stevenson, 2022 ). They call for a particular emphasis on ethical decision-making; while AI systems make analytic, solution-oriented choices based on algorithms, humans are able to make altruistic and empathetic decisions (Karakuş et al., 2025 ; Stevenson, 2022 ), taking into account “complex and emotional elements such as values, ethical principles, and moral standards” (Merzifonluoglu & Gunes, 2025 , p. 12). 2.2.2 Human Flourishing Education and AI: Developing new Models for the Future With the advent of advanced AI systems, some established models in politics, economics and society are increasingly prone to fraud. Thus, on the one hand, AI systems urge us to develop new societal, economic and organizational models and on the other hand, they can support the development of new models and structures in order to enable Human Flourishing (OECD, 2024b ). Within the realm of education, increasingly human-like AI systems are challenging human agency and accountability in teaching and learning (Darvishi et al., 2024 ; Merzifonluoglu & Gunes, 2025 ; Miao & Cukurova, 2024 ) and are reshaping teacher-student interactions (Xu & Ouyang, 2022 ). Concerns are raised about generative AI systems endangering the integrity in exams and homework (Gao et al., 2025 ), thereby undermining established educational practices. These changes suggest the need for new models within the educational landscape (Gao et al., 2025 ; Xu & Ouyang, 2022 ). Teachers are primary actors in educational processes, mediating their students’ use of AI (Miao & Cukurova, 2024 ); there is thus a particular need to reconsider teachers’ roles and competencies in the light of (generative) AI technologies (Feldman-Maggor et al., 2025; Fock et al., 2025 ; Miao & Cukurova, 2024 ; Mishra et al., 2023 ; Ng et al., 2023 ). Pivotal to these undertakings is that new models and frameworks do not discriminate against certain groups (OECD, 2024b ). This poses significant challenges as AI system are prone to amplifying biases in relation to gender, ethnicity, minorities, and disadvantaged groups inherent in their training data (Navigli et al., 2023 ), posing a threat, for example, to the validity automated assessment (Aloisi, 2023 ). Conversely, AI systems can also serve to reduce educational inequalities by catering to students with special needs (Kasneci et al., 2023 ; Yang et al., 2024 ) and learners in deprived regions with little resources (Butgereit & Van Staden, 2023). Thereby, AI may not only be a motivator for changing educational practices, but it could also assist teachers in implementing new models which recognize each student in their individual path towards a flourishing life. 2.2.3 Human Flourishing Education and AI: Restoring Meaning to Individual Lives More than prior technologies, AI systems mimic human behavior and decision-making. This poses the fundamental question of distinctly humans characteristics and whether AI systems will increasingly impair human autonomous decision-making and identity (Korseberg & Elken, 2025 ; Miao & Cukurova, 2024 ; OECD, 2024b ), thereby potentially impacting their ability to lead a flourishing life. This possible loss of purpose and identity in life is contrasted by the potential of AI to render human work more meaningful by allowing humans to focus on more meaningful activities (OECD, 2024b ). This fundamental ambivalence is reflected in the educational domain: fears of teacher replacement (Shankar et al., 2025) stand opposed to hopes for augmentation of teacher capabilities (Holstein & Aleven, 2022 ); concerns about overreliance on AI systems (Darvishi et al., 2024 ) are contrasted to the opportunity to focus on more complex meaningful activities in class (Siller et al., 2025 ). To ensure that AI is used to the benefit of students, teachers and the society at large, a human-centered approach to AI is essential (Bulathwela et al., 2024 ; Miao & Holmes, 2023 ), which “ensur[es] human agency and human accountability, and understanding AI’s societal impact and implications for citizenship in the era of AI” (Miao & Cukurova, 2024 , p. 23). The development of AI systems should not be centered on the technically possible but on the educationally promising (Barno et al., 2024). Therefore, customized AI tools need to be evaluated before application (Miao & Cukurova, 2024 ) to make sure they “function effectively while aligning with ethical standards and pedagogical goals” (Virvou & Tsihrintzis, 2024 , p. 2). 2.3 STEM Education for Human Flourishing Human Flourishing should be seen as an overarching goal rather than a goal of individual lessons or subjects (OECD, 2023 ). However, (formal) education is usually structured in lessons where technology is employed to facilitate the teaching and learning of the respective discipline and reach disciplinary goals (Ergras et al., 2022 ). We therefore narrow our focus to specific disciplines, namely to STEM education. STEM Education faces severe challenges which might affect student flourishing in the present and the future. Large-scale international student assessment studies have recently shown that 46% of the 8th grade students do not like learning mathematics and 55% of the 8th grade students report do not feel confident in mathematics (Von Davier et al., 2024 ). Similar numbers can be found for physics and chemistry. Moreover, mathematics anxiety has risen in most countries over the past 10 years (OECD, 2024a ). At the same time, STEM disciplines might be key to ensuring Human Flourishing in future societies. Human Flourishing requires intertwining scientific and moral reasoning, drawing on both adaptive problem solving and ethical decision making (Stevenson, 2022 ). This duality aligns with demands for education in STEM to nurture students’ reasonable decision-making and critical thinking and encourage active, responsible, democratic citizenship (for mathematics education, see e.g., Geiger et al., 2023 ; Siller et al., 2024 ; for science education, see e.g., Yacoubian, 2018 ). Thereby STEM education could play a key role in enabling students to contribute to Human Flourishing for their communities and to fair and sustainable futures. In particular, knowledge from diverse STEM disciplines is crucial to understanding (AI) technologies and their societal implications (Siller et al., 2025 ). Su ( 2020 ) adds to these potentials specifically for mathematics education by arguing that mathematics cultivates virtues connected to Human Flourishing such as play, truth, beauty and justice that are fundamentally rooted in human identity. This situation – challenges to Human Flourishing in STEM education on the one hand and the potential of STEM education for Human Flourishing on the other – now coincides with increasingly AI-rich teaching-learning-scenarios in STEM classrooms, which bring about their own tensions, ambivalences and transformative effects (see sections 2.2.1 to 2.2.3 ). Some of them have been explored in literature reviews on AI in STEM education. 2.4 Reviews on AI in STEM Education Reviews on (generative) AI in STEM education differ in various respects, addressing tensions, ambivalences and transformative effects of AI in STEM education from different angles. For example, Roberts and Mohamed ( 2024 ) and Humble ( 2024 ) focus conduct a SWOT analysis, which identifies s trengths, w eaknesses, o pportunities and t hreats of generative AI in computer science education. Strengths and weaknesses pertain to the characteristics of the AI system, i.e. tasks the AI system is able to complete well or poorly respectively, while opportunities and threats refer to potential uses and outcomes of using AI (Roberts & Mohamed, 2024 ). SWOT analyses originally stem from management (Helms & Nixon, 2010 ) but have been used in various other disciplines, including the field of AI in education (e.g., Denecke et al., 2023 ; Farrokhnia et al., 2024 ). Harris & Kittur ( 2024 ) take a different approach and identify research themes related to AI in education, suggesting “AI Integration”, “Ethical and Academic Integrity Considerations”, “AI's Role in Personalized Learning and Assessment”, “Challenges and Limitations of AI in Education” and “Future of AI in […] Education” as key topics addressed in the research. Identifying research themes allows researchers to “gain a better understanding of the current landscape and make reasonable predictions about the field’s future trajectory” (X. Chen et al., 2024 , p. 17496). It has also been used outside of STEM education (e.g., X. Chen et al., 2024 ; Yusuf et al., 2024 ). Both approaches highlight issues related to Human Flourishing such as overreliance, ethical concerns or individualization. The reviews on AI in STEM education also differ with respect to the disciplines, AI tools and educational levels of interest. While some reviews address mathematics education (Listyaningrum et al., 2024 ; Opesemowo & Adewuyi, 2024 ; Pepin et al., 2025 ), engineering education (Chu & Lim, 2023; Filippi & Motyl, 2024 ; Harris & Kittur, 2024 ) or computer science education (Amos et al., 2025 ; Raihan et al., 2024 ; Roberts & Mohamed, 2024 ) in particular, others focus on STEM education in general (Loke et al., 2024 ). Moreover, Opesemowo and Adewuyi ( 2024 ) focus on AI in general, Roberts and Mohamad (2024) address generative AI in particular and Pepin et al. ( 2025 ) take an even narrower approach in targeting only ChatGPT. This limitation to generative AI or even ChatGPT is understandable, given its recent surge in public attention and the need to address the potentials and challenges unique to generative AI. Furthermore, the reviews on AI in STEM education differ in the educational levels which are considered; for instance, Harris & Kittur ( 2024 ) focus on undergraduate engineering education, Chu & Lim (2023) on engineering teacher education and Loke et al. ( 2024 ) review STEM pre-service teachers’ perceptions of AI. If the educational level is not specified but analyzed in the review, articles have found a large share of studies from higher education (Filippi & Motyl, 2024 ; Raihan et al., 2024 ). This lends to the conclusion that there might be a greater focus on higher education so far, both in the reviews and the original articles on the integration of (generative) AI in STEM education. In this review, we therefore focus particularly on secondary education and the integration of generative AI. Considering that the integration of generative AI in education is in its infancy and research articles are only starting to be published, we broaden the range of potentially eligible research by including all STEM disciplines in the review. Additionally, different from previous reviews, we derive implications for future research regarding Human Flourishing. 3. Research Questions The objective of this review is to present a comprehensive overview of the current research on generative AI in secondary STEM education and to investigate to what extent the current research can serve as a foundation for exploring Human Flourishing and generative AI in STEM education. Specifically, informed by prior reviews of generative AI in STEM education, we will address three research questions: What are the characteristics of the current research on generative AI in secondary STEM education regarding discipline, publication time, country, journal or conference type, research method, data source, participants, and tools? How prevalent are different research themes in the current research on generative AI in secondary STEM education? What are the strengths, weaknesses, opportunities and threats (SWOTs) of generative AI for teaching and learning in secondary STEM education? 4. Methodology 4.1 Search Strategies and Data Collection To address the research questions, we conduct a scoping review following the PRISMA Extension for Scoping Reviews (Tricco et al., 2018 ). A scoping review aims to „determine the scope or coverage of a body of literature on a given topic and give clear indication of the volume of literature […] as well as an overview […] of its focus” (Munn et al., 2018 , p. 2). It therefore seems particularly suited to identify general characteristics, research themes as well as SWOTs. The process of identification, screening and inclusion is shown in Fig. 1 . 4.1.1 Identification The literature search was conducted by the first author using five databases – Web of Science (WoS) Core Collection (1), ERIC (2), Teacher Reference Center (TRC) (3), ArXiv (4) and SCOPUS (5) – on October 24, 2024 ((1)–(4)) and November 05, 2024 (5). In addition to the commonly used databases for educational research – WoS, ERIC, TRC and SCOPUS – we decided to include ArXiv as a database for two reasons: First, it is a key database for technical research, and the review concerns technical topics. Second, ArXiv publishes preprints, allowing for the inclusion of the latest research results, which is crucial in the rapidly developing field of AI. In order to obtain relevant research on generative AI in secondary STEM education, the search string required each record to contain at least one AI-related term, one discipline-related term and one education-related term in its title, abstract or keywords. Alternatively, a record could contain at least one AI-related term and the term “technology education” or similar combinations 2 . The initial search string was peer-reviewed by an expert on AI in mathematics education according to the PRESS guidelines (McGowan et al., 2016 ). The review resulted in the following final search string for all databases: (ChatGPT OR “Chat-GPT” OR LLM* OR “large language model*” OR “gen* AI” OR GenAI OR “Generative Artificial Intelligence”) AND (((math* OR STEM OR physics OR “computer science” OR science OR chemistry OR biology OR engineering) AND (education OR school OR college OR universit* OR “professional development” OR teach* OR student* OR learn* OR curricul*)) OR “technology education” OR “technology teach*” OR “technology student*” OR “technology curricul*”) The specifics for individual databases shown in Table 1 limit the results to English, peer-reviewed records (if allowed as a filter by the respective database) that contain the search string in their title or abstract (or keywords). We do not specify the date of publication; even though generative AI has become more present in the public debate since the release of ChatGPT in 2022, it had existed before, and earlier literature could also inform current and future research. Table 1 Specific search settings in databases Database Specific search options WoS Search in Topic and Language English ERIC Search in Abstract ( AB ) and Title ( TI ) Limiters – peer reviewed; Language: English; Expanders – Apply equivalent subjects; Search modes – Find all my search terms TRC Search in Abstract ( AB ) and Title ( TI ) Filter – Peer-Reviewed SCOPUS Search in TITLE-ABS-KEY and Language English Due to limitations in the Advanced Search mode in ArXiv, the ArXiv API was accessed via a Python (3.8) script to search for all combinations of AI-related, discipline-related and education-related terms. The Python script was used to obtain the ArXiv-IDs of each record, remove duplicates within ArXiv (i.e. not between different databases), retrieve relevant information on the records from ArXiv (title, authors, year, abstract, journal, URL) and save them to a bixtex-file. Citavi7 and Excel-Microsoft365 were used for managing and analyzing the records. The identification process led to a total of 6180 records across all five databases (see Fig. 1 ) which resulted in 4731 records after removing duplicates. 4.1.2 Screening and Inclusion The 4731 records identified after removing duplicates were screened for eligibility based on title, abstract and keywords according to the inclusion criteria in Table 2 . Apart from journal articles, we include conference publications to capture results that are not (yet) published in journals. Preprints allow us to capture the latest research results, which is highly relevant in the rapidly changing field of AI. We further screen the titles, abstracts and keywords for alignment with the three key components of the research questions, i.e. generative AI, secondary education and STEM disciplines. Table 2 Inclusion criteria Inclusion criteria Written in English Journal article, conference contribution or preprint Focusses on generative AI Focusses on secondary education Focusses on one or more STEM disciplines or STEM as a whole The screening process resulted in 243 eligible records for full-text screening. Based on the full-text reports, 183 records were identified as eligible for qualitative analysis according to the inclusion criteria in Table 2 . 4.2 Data Analysis 4.2.1 (RQ1) General Characteristics For answering the first research question, we categorized the studies according to various dimensions based on deductive category systems for each dimension: Discipline. The discipline addressed in the publications is categorized according to the different STEM disciplines (Science, Technology, Engineering, Mathematics), where Science is further divided into Physics, Chemistry, Biology and Earth Science if indicated in the publication. Moreover, we consider a category “STEM” if no further subdivision is indicated in the publication (e.g. interviews with STEM teachers). Time. As the publications are expected to be concentrated in 2023 and 2024, not only the year, but also the month of publication was identified to allow for a more nuanced analysis. Journal articles were categorized according to the month and year of their publication, conference publication according to the month and year of the conference and preprints according to the month and year of the latest upload on the respective platform (e.g., ArXiv). Country. The publications were categorized according to the countries of the affiliations of all authors. Journal or conference type. For journal and conference publications, seven types of journal or conference types were considered based on the self-description of the journal or conference: (1) focusing on education in one STEM discipline, (2) focusing on education in several STEM disciplines, (3) focusing on education in general, (4) focusing on general education and technology, (5) focusing on technology, and (6) other. Note that journals addressing one or several STEM disciplines and technology, are categorized as (1) or (2) and not as (4). Publication Type and Research method We identify the research method for original research articles as qualitative, quantitative or mixed or multi methods. We further categorize the other publications as theoretical analysis, review, editorial or position paper. Moreover, we categorize publications that do not describe the research method as reports. Data Source and participants We distinguish the data sources of original research articles as interview, survey, questionnaire, AI-generated content, test, observation, audio/video recording, student work, log data (excluding chatlogs, which are categorized as 4 + 8) and measurements. In the case of human participants, we distinguish between students, teachers, pre-service teachers (PSTs) and others. Tools We differentiate between standard generative AI systems (e.g., ChatGPT or Gemini) and non-standard AI systems, e.g. self-developed, fine-tuned tools or retrieval augmented generation 3 (RAG) enhanced systems. The complete coding scheme is provided in the supplementary material. 4.2.2 (RQ2) Research Themes We conducted a structuring qualitative content analysis of the full-text publications (Kuckartz & Rädiker, 2023 ) to address the second research question. Our deductive approach included the development of main categories, which we used to inductively develop subcategories for on the most frequent main categories (Kuckartz, 2019 ). The main categories were developed by collecting various articles that investigate research themes for AI in education (see section 2.4 ) and synthesizing the different category systems from these contributions. The synthesized category system included nine categories: Theoretical frameworks Evolution, future directions, trends & philosophical considerations Pros & cons Tool development Performance Ethical considerations & academic integrity Awareness, attitude & acceptance Application & impact Educational research The category systems including category descriptions can be found in the supplementary material. The full texts were considered as a coding unit. In principle, we allowed double coding, but only if the paper was divided into separate parts with different research goals; otherwise, we assigned only one research theme according to the dominant theme of the publication. 4.2.3 (RQ3) SWOT Analysis For addressing the third research question, we conducted a SWOT analysis of the records using structuring qualitative content analysis (Kuckartz & Rädiker, 2023 ). The four components of SWOT – strengths, weaknesses, opportunities and threats – were used as main categories developed by deductive category formation. A precise description of these categories – similar to prior research (see section 2.4 ) – is shown in Table 3 . Subcategories were developed inductively to ensure that the categories can be specific to the STEM disciplines. Table 3 Main categories of SWOT analysis Category Description Strength Property of generative AI systems that is conducive to the teaching and/or learning of the respective STEM discipline(s) in secondary education Weakness Property of generative AI systems that is obstructive to the teaching and/or learning of the respective STEM discipline(s) in secondary education Opportunity Influence of generative AI systems on teachers and students that fosters the teaching and/or learning of the respective STEM discipline(s) in secondary education Threat Influence of generative AI systems on teachers and students that inhibits the teaching and/or learning of the respective STEM discipline(s) in secondary education In order to obtain evidence-based results, we included only empirical research in the SWOT analysis. Sentences or parts of sentences in the abstract and conclusion of each publication were used as coding units. Abstract and conclusion are meant to summarize the findings and identify their main implications and are therefore particularly suitable for the identification of SWOTs. Instead, coding sentence-wise in the whole publication would pose the risk of losing sight of the context and overrepresenting individual statements instead of the main message of the paper. The analysis was conducted using MaxQDA24. The paraphrases of all passages in a main category were collected. Using paraphrases for inductive category development is inspired by Mayring’s ( 2014 ) summarizing content analysis and applied here for two reasons: first, the large number of passages makes verbatim collection difficult, and second, paraphrasing incorporates contextual information (e.g. about the discipline referred to). Inductive development of subcategories was applied for each of the four main categories based on the paraphrases. 5. Results 5.1 Characteristics of the current research on AI in secondary STEM education Regarding the distribution of publications across disciplines, mathematics is the most frequently covered discipline (39%), followed by physics (24%) and science (19%). All other disciplines as well as STEM as a whole are addressed in 10% or less of the publications (see Fig. 2 ). Moreover, biology, chemistry and earth science are addressed in conjunction with other disciplines in the majority of publications where they occur. Regarding the month and date of publication, an overall increase can be noted (see Fig. 3 ). There is a peak in July 2024, which can be mainly attributed to several conferences on AI in education at this time. The downwards trend after July 2024 should not be overinterpreted, as a drop from July to August also occurred in 2023 and for Oktober 2024, not all results are included due to a retrieval date for this review. Furthermore, it can be noted that one publication was published in May 2021, while all other papers are published after 2023, i.e. after the release of ChatGPT. Note that four publications were not included in Fig. 3 , as they used to be preprints at the time of identification of records but had been published at the time of full-text retrieval. The geographical distribution reveals that the United States are by far the most active country with the most contributions, accounting for 30% of the global publications, followed by China (8%), Germany (8%) and the United Kingdom (5%) and Brazil (5%). Overall, 73% of the affiliations are situated in the Global North and only 27% in the Global South. The distribution across different disciplines shows that the concentration in the Global North is particularly strong for technology education and science education and less than average for all other disciplines. The concentration in the Global North is even more pronounced for conference publications, where 88% are contributions from the Global North and 48% from the United States. Conversely, for journal articles, the geographical distribution is a bit more levelled out, with 15% from the United States, followed by Germany (13%), China (9%), Brazil (6%), Australia (6%) and South Africa (4%). For the journal or conference type, it should first be noted that 14% of the papers are preprints and thus not considered here. Of the remaining contributions (33% conference publications, 53% journal articles), 17% are published in journals or at conferences targeting one STEM discipline, 14% in journals or on conferences focusing on several STEM disciplines. Another 10% of journals and conferences are dedicated to education in general. More than half of journals and conferences either focus on technology (18%) or on education in conjunction with technology (34%). The remaining 7% do not belong to any of these categories (see Fig. 4 ). For individual disciplines, papers from the sciences are more frequently published in STEM-specific journals and conferences, while this is rare for mathematics. Mathematics papers are predominantly published in journals and conferences on education and technology. Regarding the methodology, we consider research method, data source, and participants. Out of the 183 publications, 76% present original research articles, which include full data collection, analysis and presentation of results. They are almost equally split across qualitative (33%), quantitative (30%) and mixed/multi methods (37%) research designs. 16% of the publications are reports without a description of the methodology, describing for example a teaching unit or the development process of an AI tool. The remaining publications are reviews (2%), editorials (1%), position papers (1%) and theoretical analyses (4%) (see Fig. 5 ). Concerning individual disciplines, the share of original research is particularly high for biology, earth science, technology and mathematics. Moreover, mathematics, physics, and science reflect the almost equal split across qualitative, quantitative and mixed-methods, while chemistry is dominated by qualitative research and STEM is dominated by quantitative and mixed-methods designs. The number of reports is particularly high in chemistry and rather low in mathematics. Regarding the data source (see Fig. 6 ), half of the original research publications collect, analyze and present AI-generated content. For papers investigating human participants, using surveys, interviews and students work is most common, often in conjunction with each other or with AI-generated content. Moreover, the most commonly assessed group of human participants are students, followed by teachers, pre-service teachers and others (e.g. parents, teacher educators). With regards to the tools, 9% of the publications use (sometimes among others) fine-tuned or self-developed AI systems or integrate retrieval augmented generation (RAG). The remaining 91% resort to existing AI systems, predominantly to ChatGPT. 5.2 Research Themes Figure 7 : Research themes overall (black) and divided by discipline (colored); bars are higher for disciplines because publications addressing multiple disciplines are counted multiple times for the respective research theme The results show that performance is the most frequent research theme (35%), followed by tool development (22%) and application & impact (19%) of AI tools. All other themes are rather rare. The different themes are approximately equally distributed across the disciplines. In particular, performance is the most frequent category for all disciplines except for the three infrequent ones ( technology , engineering and STEM ). Some disciplines are not represented across all research themes: Notably, the most frequent discipline – mathematics – does not appear for ethical considerations & academic integrity . Moreover, mathematics is represented in the application & impact theme almost as frequently as physics, although the overall number of mathematics-related publications is much higher. Interestingly, for technology education, application & impact is the most common theme. We take a closer look at the research themes by (1) presenting subcategorizations of the three largest research themes and (2) considering the development of research themes over time, in relation to data sources and participants as well as the tools used across the research themes. Figure 8 shows that the kind of performance assessed ranges from solving mathematical tasks – in competition problems (e.g., Teegavarapu & Sanghvi, 2023), official exams (e.g., Dao & Le, 2023), and other questions (e.g., Schorcht, Buchholtz, et al., 2024) – to engaging in pedagogical considerations by planning lessons (e.g., Hu et al., 2024), tutoring and dialogs (e.g., Gregorcic & Pendrill, 2023), or generating feedback and assessment (e.g., Bewersdorff et al., 2023). Similar categories can be identified for tool development , which shows that apart from assessing the performance of generative AI in pedagogical activities, there are efforts to improve the performance of generative AI in these areas. Moreover, almost a quarter of the publications in the tool development category is concerned with the compilation of datasets, for example for multimodal mathematical reasoning (e.g., Z.-Z. Li et al., 2024) or high school physics questions (e.g., Anand et al., 2023). For application & impact , it can be found that 33% of the publications focus on the impact of using generative AI on cognitive variables (e.g., Dasari et al., 2024; Ghazali et al., 2024; T. Li et al., 2024) such as performance in STEM disciplines or cognitive load. Another 8% address effects on non-cognitive variables (e.g., intrinsic motivation or self-confidence) and 17% focus on both (e.g., Canonigo, 2024; Lu et al., 2024). The remaining publications describe the behavior of participants while engaging in activities with AI (e.g., the behavior of students working with GeoGebra using ChatGPT in Yunianto et al., 2024) (17%), describe an activity about dealing with AI and assess student performance in it (e.g. with AI-generated science texts in Cheung et al., 2024) (8%) or solely describe teaching or learning activities in which generative AI (Pavlova, 2024) (17%). We further consider the development of the research themes over time by three-month units to avoid rapid fluctuations. Figure 9 (left) shows the prevalence of research on the performance of AI tools at all times, as well as a rapid increase of research on tool development since the end of 2023. In the beginning of 2024, the tool development category surpasses the application & impact category. For the remaining categories, the total number of publications is too low to draw reasonable conclusions across time. Relating the participants to research themes reveals some noticeable patterns (see Fig. 9 , right). Unsurprisingly, most publications on tool development and performance of AI tools do not involve humans. However, if tools are tested with humans, participants are mostly students. Moreover, for the awareness, attitude & acceptance category, participants are mostly teachers. Conversely, when impact and application of AI tools is assessed, this is very rarely done with teachers – the most frequent participant groups are students, followed by pre-service teachers. Regarding the use of standard AI tools versus non-standard AI tools (i.e. self-developed tools, fine-tuned tools or tools using RAG), it can be noted that non-standard tools are used only in 9% of the publications, mostly for the research themes tool development and performance . Publications in other research themes (almost) exclusively use standard tools. 5.3 SWOT Analysis The SWOT analysis summarizes the strengths, weaknesses, opportunities and threats of generative AI in secondary STEM education. The results are presented in Fig. 10 . As elaborating on each factor for all STEM disciplines would go beyond the scope of this review, we focus on mathematics as the most prominent discipline in the review and at times include examples from other disciplines if only few or no mathematical examples are available. 5.3.1 Strengths Generative AI systems are able to accurately solve STEM tasks in official/national exams such as the NAEP mathematics questions (Wei, 2024). They excel particularly at basic (Dao & Le, 2023), knowledge-level mathematics questions (Guler et al., 2024) and basic mathematical modelling tasks (Spreitzer et al., 2024). Strengths can be found especially in the areas of algebra (Teegavarapu & Sanghvi, 2023), numbers (Dao & Le, 2023; Teegavarapu & Sanghvi, 2023; Wei, 2024) and probability and statistics (Vankúš, 2024). Beyond solving tasks, generative AI systems can provide explanations of solutions (Daher & Gierdien, 2024; Ergene & Ergene, 2025) and offer explanations for their scoring of student answers (Lee et al., 2024). Moreover, generative AI systems stand out for their ability to provide mathematics tutoring in various languages (Butgereit & Van Staden, 2023). Regarding educational materials, generative AI systems can create contextually relevant questions (van Pham et al., 2024), adapt mathematics questions to diverse ability levels (Rouzegar & Makrehchi, 2024) and effectively tailor the language of mathematics tasks to different learners (Norberg et al., 2024). They further stand out for their ability to create clear and organized lesson plans with well-defined instructional objectives, organization, methods and strategies, particularly in the areas of functions and statistics (Hu et al., 2024). Generative AI systems can analyze discourse in mathematics classrooms similar to human coders (Long et al., 2024) and imitate mathematics students (Drushlyak et al., 2024; Lu et al., 2024; Rouzegar & Makrehchi, 2024). An overarching strength of generative AI systems is their customizability for educational purposes: This can improve the generation of questions (R. Li et al., 2024), distractors and error labels (Fernandez et al., 2024) in mathematics multiple-choice questions, enhance the performance in mathematics question answering (Levonian et al., 2023; Zhang et al., 2024) and lead to more effective simulations of students (Jin et al., 2024; Sonkar et al., 2024 ) – for example in collaborative mathematical modelling (Yue et al., 2024) – as well as to more accurate scoring and evaluation (Latif & Zhai, 2024; Nicula et al., 2023). Other strengths of generative AI for STEM education pertain to the avoidance of bias in some cases (Cooper & Tang, 2024; Kunz & Steffen, 2024) and to their large knowledge base, providing interdisciplinary knowledge (dos Santos, 2023). 5.3.2 Weaknesses A major weakness of generative AI systems is their tendency to produce wrong information and make mistakes (Butgereit & Van Staden, 2023; Daher & Gierdien, 2024; Ergene & Ergene, 2025; Guler et al., 2024; Prihar et al., 2023; Ribeiro et al., 2024; Spreitzer et al., 2024; Taani & Alabidi, 2024). Difficulties arise particularly for complex problems (Dao & Le, 2023; Teegavarapu & Sanghvi, 2023; Wei, 2024) and problems that require nuanced understanding (Spreitzer et al., 2024), formal reasoning (Parra et al., 2024) or induction (Dasari et al., 2024). Generative AI systems struggle with spatial reasoning and geometry (Dao & Le, 2023; Guler et al., 2024; Wardat et al., 2023; Wei, 2024), calculus (Dao & Le, 2023) as well as combinatorics and probability (Guler et al., 2024; Teegavarapu & Sanghvi, 2023). Problems in geometrical tasks align with a general difficulty of generative AI systems with graphical input and output – they struggle for example in visual proofs (Schorcht, Baumanns, et al., 2024) and multimodal mathematics reasoning (Z.-Z. Li et al., 2024). Also, performance may be worse for non-English mathematics tasks (Dao & Le, 2023; Parra et al., 2024). Apart from committing mathematical errors, generative AI systems also struggle with anticipating (Feng et al., 2024), understanding (McNichols et al., 2024) and categorizing (Yan et al., 2024) (common) mathematical errors. They further show difficulties in identifying student reasoning for flawed solutions (Liu et al., 2024) and differentiating mathematical errors from student insecurities (Kakarla et al., 2024). As generative AI systems are based on huge amounts of text but not on real-world experiences, they may lack real-world knowledge when creating mathematics lesson plans, for example on the curriculum, mathematical culture and suitable technology integration (Baytak, 2024; Egara & Mosimege, 2024; Hu et al., 2024). Sometimes, the AI-generated educational materials – distractors (Feng et al., 2024; McNichols et al., 2023), guidance prompts (Dasari et al., 2024), generated questions (Vankúš, 2024) and hints (Gattupalli et al., 2023) – do not keep up with human ones or are less preferred by experts. Other weaknesses of generative AI for STEM education concern the issue of bias, for example concerning STEM job recommendations regarding gender (Due et al., 2024) or the negative perception of mathematics and STEM (Abramski et al., 2023). 5.3.3 Opportunities We differentiate between opportunities for students, for teachers and for pre-service teachers. Generative AI systems can assist students in learning and understanding mathematics (Canonigo, 2024; Daher & Gierdien, 2024; Egara & Mosimege, 2024; Schorcht, Buchholtz, et al., 2024; Yunianto et al., 2024). For example, they can translate mathematical language to natural language (Ribeiro et al., 2024) and integrate various modalities (Schorcht, Baumanns, et al., 2024), leading to an increase in (conceptual) mathematical understanding (Canonigo, 2024; Egara & Mosimege, 2024). Similarly, generative AI systems can help students solve (basic) mathematics tasks and thereby lead to increased student performance (Henkel, Horne-Robinson, Kozhakhmetova, et al., 2024; Norberg et al., 2024; Rouzegar & Makrehchi, 2024). Moreover, generative AI systems can promote self-directed, (inter)active learning, for example in mathematics flipped classroom settings (Pavlova, 2024). The use of generative AI can also contribute to the students’ development of critical thinking and reflection skills, for example when students categorize errors of AI systems in geometry (Parra et al., 2024) or analyze mathematical mistakes made by these systems (Bellettini et al., 2023). Generative AI can further promote problem-solving activities and computational thinking when students debug AI generated GeoGebra commands (Yunianto et al., 2024) or create mathematical visualizations using generative AI (Schorcht, Baumanns, et al., 2024). If students solve mathematical tasks collaboratively with an AI system (Daher & Gierdien, 2024) or engage in AI-guided reward-based learning paths (Singh et al., 2024), these systems have the potential to enhance engagement and collaboration in mathematics. Furthermore, they promote inclusivity, democracy and equality, for example by assisting teachers in creating specialized math practice for disabled students (Lin & Riccomini, 2024) and making mathematics education accessible in lower and lower-middle income countries (Henkel, Horne-Robinson, Kozhakhmetova, et al., 2024). All in all, the use of generative AI systems in mathematics education can positively influence affective variables by increasing self-efficacy, confidence (Canonigo, 2024) and self-assurance (Yunianto et al., 2024) in mathematics. Generative AI can assist teachers with lesson planning (Baytak, 2024; Hashem et al., 2023; Taani & Alabidi, 2024) and serve as reference for teachers creating mathematics explanations (Prihar et al., 2023) and tips (Jia et al., 2024). Generative AI can thereby reduce teachers’ workload and increase their teaching efficiency (Egara & Mosimege, 2024; Hashem et al., 2023; Hu et al., 2024). Apart from the creation of educational resources, generative AI systems can help teachers with assessment and feedback. They may analyze students’ responses and weaknesses in mathematics and science (Taani & Alabidi, 2024) and encourage open-ended, process-oriented assessment formats (Henkel, Horne-Robinson, Dyshel, et al., 2024). Moreover, generative AI systems allow teachers to analyze and practice their own teaching by generating feedback for their teaching (Barno et al., 2024), particularly regarding socio-emotional learning (Han et al., 2024). Pre-service teachers are on the one hand themselves learners of mathematical content in university courses where generative AI can assist them in completing tasks and essays (Dasari et al., 2024; Vankúš, 2024), leading to higher performance if used in conjunction with a human teacher (Dasari et al., 2024). On the other hand, pre-service teachers need to develop teaching skills. Using generative AI systems, they can practice finding and arguing mathematical errors (Drushlyak et al., 2024), which may lead to higher teacher self-efficacy (Lu et al., 2024), increased critical thinking (Drushlyak et al., 2024) and increased higher order thinking (Lu et al., 2024). 5.3.4 Threats We differentiate between threats for students, for teachers and for pre-service teachers. Using generative AI in mathematics classrooms can hinder students’ learning and understanding of mathematics as the system may give immediate answers (Singh et al., 2024) and reinforce or even introduce misconceptions (Daher & Gierdien, 2024; Parra et al., 2024). Generative AI systems may further impede the development of critical thinking and reflection (Dasari et al., 2024; Shankar et al., 2025), independent problem-solving and independent analysis and evaluation of information as students may overly rely on AI systems (Shankar et al., 2025). Other threats of generative AI in STEM education pertain to the reinforcement of stereotypes (Cooper & Tang, 2024; Due et al., 2024), ethics and data security (Latif & Zhai, 2024; Shankar et al., 2025) as well as dishonesty and misbehavior of students (Garofalo & Farenga, 2025; Yeadon & Hardy, 2024). For teachers, the potential marginalization or replacement of human teachers’ involvement in the learning process and of the subtleties of their expertise presents a major threat (Barno et al., 2024; Vasconcelos & Dos Santos, 2023). The adoption of generative AI systems may disadvantage teachers with less digital competencies (Shankar et al., 2025), replace teachers as well as traditional teaching and grading by automation (Latif & Zhai, 2024) and focus too much on the technologically possible and not on the educationally promising in mathematics teacher professionalization (Barno et al., 2024). Pre-service teachers may suffer from decreased performance in mathematics when only learning with generative AI systems (i.e. without a teacher) (Dasari et al., 2024). They may also experience a lower quality of educational materials regarding the clarity of tasks in physics (Küchemann et al., 2023) and the design of STEM teaching units (Z. Li & Ironsi, 2024). 6. Discussion Education for Human Flourishing strives to enable all students to live up to their potential in life by fostering a sense of purpose, meaning and autonomous decision-making. Recent technological developments call for an exploration of the role of (generative) AI regarding Human Flourishing in educational settings including secondary STEM education. To this end, this article has reviewed the current research on generative AI in secondary STEM education considering general characteristics, research themes and SWOTs. Based on these results, we are going to discuss to what extent the current literature provides a foundations for exploring the interaction of Human Flourishing and generative AI in secondary STEM education, discussing the findings under the three ideas proposed by the OECD (2024) – Broadening human capabilities , developing new models for the future , and restoring meaning to individual lives – each of them in relation to secondary STEM education. Based on these elaborations, we outline potentials for future research. 6.1 Broadening Human Capabilities in STEM Education Increasingly advanced AI systems underline the need for a shift from mostly cognitive educational goals to socio- and meta-cognitive objectives in STEM education, focusing on a holistic view of the student (OECD, 2024b ). However, the findings of this review show that the effect of using generative AI on cognitive variables is investigated more frequently than on non-cognitive variables. Despite the apparent emphasis on cognitive aspects, the SWOT analysis draws a more holistic picture, revealing opportunities for learners’ motivation, confidence and self-efficacy, their engagement and collaboration in STEM disciplines as well as the advancement of inclusivity and equality in STEM teaching and learning. These opportunities provide initial evidence that generative AI might be helpful to address challenges to Human Flourishing in STEM education, such as low student confidence and increasing anxiety (OECD, 2024a ; Von Davier et al., 2024 ). They further highlight that AI could provide an opportunity to foster capabilities that enable students to uphold circumstances that allow flourishing throughout society, such as critical thinking and active engagement as responsible citizens (Geiger et al., 2023 ; Yacoubian, 2018 ). There is a need for further research that comprehensively examines the impact of generative AI on students, including metacognitive and socio-emotional factors, to strengthen initial research efforts regarding the associated opportunities. The need for broader and more holistic human capabilities places increased emphasis on human ethical decision-making as opposed to automatic choices (Karakuş et al., 2025 ; Stevenson, 2022 ). However, current research shows that there is a limited body of literature addressing ethical considerations and academic integrity, particularly in mathematics education. However, the SWOT analysis shows that various publications mention the threat of misbehavior by students, ethics and data security as well as the reinforcement of stereotypes through generative AI systems. There is a need to explore ways to mitigate these issues in STEM education in order to ensure that the technology promotes inclusivity and democratization (Ergras et al., 2022 ) and fosters ethically sound educational practices (Virvou & Tsihrintzis, 2024 ). There is a need for publications that focus on ethical consideration in STEM education in depth. By leveraging empirically-based opportunities and mitigating threats, STEM education could play a significant role in fostering Human Flourishing as it provides opportunities to intertwine ethical and scientific reasoning, fostering adaptive problem solving in conjunction with ethical decision-making (OECD, 2024b ; Stevenson, 2022 ). Given this pivotal role of STEM, our review has included only publications concerning STEM education in some regard. The findings revealed that less than a third of included journal articles and conference contributions are published in STEM education journals or conferences; in turn, almost half of the papers are published in general education journals or conferences, mostly in conjunction with technology. This lends to the conclusion that research on generative AI in STEM education focuses more on pedagogy and technology than on the STEM disciplines themselves. This could complicate drawing targeted conclusions for STEM teaching and learning. Given the unique potentials and challenges for Human Flourishing in the STEM disciplines, future research should focus more strongly on discipline-specific applications and implications of generative AI in STEM education in order to derive targeted decisions for STEM classrooms. 6.2 Developing New Models for the Future of STEM Education Enhanced capabilities of generative AI systems present challenges to existing – organizational, political and societal – models in education, calling for new approaches and solutions, particularly concerning teacher education (Feldman-Maggor et al., 2025; Miao & Cukurova, 2024 ; Ng et al., 2023 ). However, this review identified only very few publications that focus on theoretical frameworks for teaching and learning with generative AI in secondary STEM education. While there are diverse overarching frameworks, for example on AI-specific TPACK (Mishra et al., 2023 ), they do not specifically focus on STEM. This reflects the tendency for publications to focus rather generally on education (see section 6.1 ). At the same time, AI provides an opportunity to assist the development of new models (OECD, 2024b ) by helping STEM teachers improve their own teaching through advanced discourse analysis tools and AI-generated suggestions for improvements (see SWOT analysis). Future research should suggest STEM-specific frameworks on the integration of AI in education and also empirically validate them in order to develop new solutions in an age of generative AI. When new solutions are developed, it is crucial that they do not discriminate against minorities and ensure that also marginalized voices are heard (OECD, 2023 ). This is particularly relevant with regard to Human Flourishing as different cultures may conceptualize a flourishing life differently (Curren et al., 2024 ; de Ruyter et al., 2022 ). However, we found a concentration of publications in the Global North, especially in the United States. This is particularly unfortunate in the field of generative AI, as generative AI systems demonstrate linguistic versatility and a potential to promote inclusivity, democracy and equality in STEM education (see SWOT analysis). Research situated in the Global South does indeed show promising results of AI-assisted mathematics tutoring in under-resourced contexts (Butgereit & Van Staden, 2023). A more diverse authorship could also decrease the threat of bias and reduce stereotypes in generative AI systems, as more diverse data could be used for future training. At least, there seems to be some more balance regarding the geographical distribution for journal articles, with contributions from Brazil, South Africa and China in particular. The global research community should foster more equal participation, particularly for conferences, in order to explore how generative AI in STEM education can enhance Human Flourishing everywhere. The need for new models, particularly approaches concerning teacher-AI-collaboration in STEM education, raises the question whether a SWOT analysis could be used in this regard. The potential of SWOT analysis consists in providing a comprehensive overview of internal and external factors influencing the successful adoption of generative AI in education. Considering the complexity of the adoption of new technology and the interconnectedness of internal and external factors, it is not sufficient to regard the four components of SWOT in isolation. Moreover, SWOT analyses are not an end in themselves, but a starting point for strategic planning (Helms & Nixon, 2010 ). We therefore focus on how teachers can harness the opportunities of generative AI to assist them with the creation of educational materials and the assessment and feedback in STEM education, thereby reducing teacher workload, while mitigating the threat of teacher replacement and the lack of involvement in the learning process (see Fig. 11 ). The current strengths of generative AI provide an indication of what teachers could use generative AI systems for, while the weaknesses need to be compensated by the teacher. Teachers could use generative AI for finding suitable contexts for tasks and easily generate examples, while it is up to the teacher to check the correctness, especially for complex, counterintuitive reasoning tasks and tasks involving graphical input and output. Generative AI may be able to provide clear structure for lesson plans or other teaching activities, to which teachers then add their real-world knowledge. Generative AI can be used to further adapt the materials to different students’ needs and change formulations or even language. Teachers should watch out that no biases are perpetuated by the AI system and intervene if necessary. Regarding assessment and feedback, generative AI, especially when customized for educational purposes, may provide a starting point for scoring, excelling at providing explanations for its decisions. Teachers may then check the scoring, watching out for typical student errors and misconceptions. It is crucial to highlight that the presented strategy is not fixed, but subject to dynamic changes as generative AI systems evolve and new research results become available; for instance, current weaknesses that need to be compensated by the teacher, might become strengths in the future (or vice versa), demanding different teacher reactions. Future research could implement the suggested strategy in STEM teacher professionalization and evaluate whether it actually reduces teacher workload and increases efficiency, while maintaining the central role of the teachers in the educational process. With regards to Human Flourishing, the suggested strategy aims to maintain teacher agency and autonomous decision-making. It selectively delegates tasks to a generative AI system while the teacher remains accountable for final choices on educational materials and student assessment. By balancing human and algorithmic decision-making (Merzifonluoglu & Gunes, 2025 ), it strives to allow teachers to enhance and not reduce the purposefulness and meaningfulness of their work (OECD, 2024b ). This anticipates the next section on restoring meaning to individual lives . 6.3 Restoring Meaning to Individual Lives in STEM Education Increasingly human-like AI systems bear the risk of challenging human identity, agency and autonomous decision-making. It is therefore crucial to adopt a human-centered mindset where the impact of generative AI on humans, the society and the environment is of primary importance. This requires empirical research on the application and impact of generative AI in teaching and learning scenarios in real STEM classrooms (Kong & Yang, 2024 ). The examination of research methods in the review shows that the vast majority of studies are empirical. However, the primary emphasis of these studies is on the evaluation of AI-generated content investigating the performance of generative AI systems in STEM education or developing and testing customized tools for STEM education. These results lend to the conclusion that the technically possible might be more strongly focused on than the pedagogically promising (Barno et al., 2024). The trend towards tool development instead of the assessment of impact and application is even slightly increasing over time, with tool development surpassing impact & application in the beginning of 2024. These results suggest an increasingly technocentric approach, which does not “put[…] the learner at the epicentre of the design of the tools” (Bulathwela et al., 2024 , p. 11). Also, newly developed or customized AI systems are rarely tested with human participants, exposing students and teachers to unexpected risks regarding their cognitive and affective development and well-being (Miao & Cukurova, 2024 ). Vice versa, the application and impact of generative AI in STEM education are never examined using fine-tuned or self-developed tools or tools using RAG. These results suggest a gap between technical development and the actual implementation. Future research should place stronger emphasis on the investigation of human application of generative AI in STEM education as well as the impact on students, teachers and pre-service teachers to adopt a human-centered approach to AI in STEM education. This should include increased efforts for cooperation between technical development and practical assessment. For the research themes that primarily involve human participants – awareness, attitude & acceptance and application & Impact – teachers constitute the largest group of participants in the theme of awareness, attitude & acceptance , while research regarding the application & impact of generative AI by teachers is largely absent. This tendency counteracts the demand that “future research should strive to study these [AI] systems in real-world contexts, with authentic tasks and relevant human experts.” (Holstein & Aleven, 2022 , p. 245f). It could also hinder a human-centered approach to AI integration in education (Bulathwela et al., 2024 ; Miao & Holmes, 2023 ). The lack of studies for application & impact involving teachers results in the lack of nuanced opportunities and threats for teachers in the SWOT analysis. Furthermore, the performance and competence of teachers is rarely assessed, in contrast to the frequent assessment of students, as evidenced by the subcategory of application & impact . However, to engage with generative AI, teacher competencies are of pivotal importance (Mishra et al., 2023 ; Ng et al., 2023 ). There is a need for research that goes beyond teachers' attitudes and opinions to investigate actual integration practices and competencies. This could facilitate the identification of empirically-based, nuanced opportunities and threats in order to ensure a human-centered approach to generative AI in STEM education. The lack of studies involving human participants might have implications for the validity of the SWOT analysis, as the opportunities and threats are sometimes based on only a few empirical studies with human participants. This could be one reason why the SWOT analysis demonstrates seemingly contradictory results: learning and understanding , problem solving and critical thinking appear both as opportunities and as threats of generative AI in STEM education. This underlines the fundamental ambivalences and tensions associated with generative AI (OECD, 2024b ). The lack of empirical evidence on the actual implementation of generative AI in STEM education might be one reason for this divergence. Some threats and opportunities of generative AI are derived from teachers’ opinions and not from an actual assessment of the impact of generative AI. Additionally, variations in the findings of different studies may result from the diverse learning scenarios being evaluated: For example, Dasari et al. (2024) not only compare a group working with AI against a traditional teaching group in mathematics but also compare an AI-only group with a group working with AI and a teacher. They find that while the AI-only group shows worse mathematics results than the traditional teaching group, the AI-plus-teacher group shows better results than the traditional teaching group. Chen & Chang (2024) compare the impact of generative AI on digital game-based learning in physics by comparing a group learning with AI and examples (e.g., of suitable prompts) against an AI-only group, finding better learning outcomes for the former. There is a necessity for more such nuanced and differentiated research conditions to explore integration methods that enhance student learning, critical thinking, and problem-solving skills, particularly considering the role of the teachers in (successful) AI-mediated educational settings. This approach could address the seemingly contradictory results of the SWOT analysis. The focus on the research themes of performance and tool development – while arguably problematic regarding a human-centered approach to AI – could serve as a starting point for future research. The evaluation of the correctness and adequacy of AI-generated content is undeniably important for its application in the classroom (Virvou & Tsihrintzis, 2024 ). A closer look has shown that subcategories of these research themes show a considerable diversity: they range from the capacity of generative AI to solve mathematics tasks, to the generation of feedback, hints, explanations and questions, to the planning of lessons and activities as well as tutoring and student simulations. While there is some overlap among these categories (e.g., answering mathematical questions may be a prerequisite for effective tutoring), they nonetheless illustrate the breadth of the research landscape. This suggests that there is only a limited number of publications within each category, which complicates drawing definitive conclusions – an issue that is exacerbated by the rapid changes of AI technologies. This also implies that the findings on strengths and weaknesses in SWOT analysis are partly based on only a few publications. Still, the strengths and weaknesses in the SWOT analysis provide initial insights into the possibilities of generative AI to enhance Human Flourishing, for example by creating personalized educational materials that meet individual student needs, and by their linguistic versatility (Kasneci et al., 2023 ). At the same time, biases and linguistic exclusion might risk inclusivity, equality and accessibility in STEM education. Future research should systematically increase the number of publications for the different subcategories of performance and tool development . An extension of research efforts within one particular subcategory could serve as a basis for an in-depth evaluation with human participants, while the evaluation of an AI system across different subcategories might allow for a more holistic picture of the educational capabilities of the system and, in a next step, of its impact on students’ and teachers’ flourishing. 7. Limitations Although we have provided a broad overview of the current literature on generative AI in STEM education, including 183 full-text publications, our review only presents the state of October 2024 and does not consider research conducted afterwards. This is clearly a limitation in such a rapidly evolving research field, particularly with regard to strengths and weaknesses, which may change with newer versions of generative AI systems. Moreover, even though we have included five common databases for educational and technical research, we cannot guarantee that some relevant research results are not listed in these databases. The inclusion of preprints in the review – while providing access to the latest research results – presents the challenge of quality assurance, especially if preprints have been published more than a year ago and have not yet been published. However, as only 15 percent of the publications are preprints, we suppose that this issue does not impede the validity of our results too much. Regarding the SWOT analysis, it should be noted that we did not indicate the version of the generative AI system used or the date of data collection for strengths and weaknesses. It could be the case that some weaknesses are no longer weaknesses of state-of-the-art AI systems today or even vice versa, that new weaknesses emerge with more advanced tools. We therefore emphasize once again that our SWOT analysis only presents a starting point for decision-making and future research and that the exact strengths and weaknesses are dynamically changing. It could serve as a reference for future research on what aspects of generative AI systems could be interesting to focus on. Future research could also track the development of the performance of generative AI systems for STEM education over time in order to derive more nuanced conclusions about strengths and weaknesses and consequently about opportunities and threats for students and teachers. Another limitation concerning that SWOT analysis is that we only examined abstracts and conclusions for strengths, weaknesses, opportunities and threats. While we aim to achieve a more representative result by this (see section 4.2.3 ), we might have missed subtleties mentioned in the publications which did not make it into the abstracts and conclusions. 8. Conclusion Education for Human Flourishing strives to enable students to engage in meaningful relationships and activities and enable them to lead a purposeful life for themselves and others. With rapid advances in (generative) AI technologies, the consideration of Human Flourishing and its entrenchment with AI has gained traction. This research has conducted a scoping review of the literature on generative AI in secondary STEM education to explore to what extent the current research provides a foundation for exploring Human Flourishing and AI in secondary STEM education. Following the PRISMA Extension for Scoping Reviews, we identified 183 eligible publications from five databases. These publications were examined with regard to their general characteristics, research themes and SWOTs and interpreted in the light of Human Flourishing. Our findings suggest that the current literature on generative AI in STEM education does not (yet) allow for an empirically-based, holistic and inclusive exploration of Human Flourishing and AI in STEM education. Despite the necessity of broadening human capabilities due to enhanced AI technology, it was shown that more publications addressed cognitive than non-cognitive variables and research on ethical considerations is underrepresented. This complicates conclusions about the holistic impact of generative AI on students in STEM education. Regarding the need to develop new models for the future to successfully address challenges posed by generative AI, the current literature shows a lack of STEM-specific theoretical frameworks. Moreover, research is concentrated on the Global North, which increases the risk of biased solutions, potentially neglecting non-Western perspectives on Human Flourishing. Concerning the restoration of meaning to human lives in the face of generative AI, it became apparent that the focus of the current research lays on the performance of generative AI systems and on tool development, investigating primarily AI-generated content instead of human participants. This approach may hinder the adoption of a human-centered mindset towards generative AI in education and might leave the strengths of AI systems identified in the SWOT analysis unleveraged. Despite these potential shortcomings of the current research, the SWOT analysis has outlined opportunities regarding the influence of generative AI on Human Flourishing in STEM education. Threats identified in the SWOT analysis also highlighted threats to Human Flourishing in STEM education, particularly ethical issues. Moreover, we have shown that a SWOT analysis might serve as a basis for new strategies of teacher-AI-interaction in a way that preserves STEM teachers’ agency, autonomy and accountability. It is based on a huge number of publications that have identified strengths and weaknesses, which offer a diverse range of empirical results. They could serve as a basis for subsequent investigations with human participants. The preliminary conclusions of this scoping review might inform future research on generative AI in STEM education in order to steer the field into a direction where Human Flourishing is at the center of educational efforts – due to, in spite of, and through generative AI. Abbreviations PST Pre-service teacher RAG Retrieval augmented generation SWOT Strengths, weaknesses, opportunities and threats Declarations Funding Not applicable as this research was not funded Author Contribution A.F. has conducted the literature search and analysis of full-text publications. A.F. and H.S.S. have drafted an initial version of the manuscript. H.S.S. has guided the theoretical conceptualization of the article and substantially contributed to manuscript revision. Both authors have read and approved the final manuscript. Data Availability A list of publications included in the scoping review as well as the coding scheme is included in the supplementary information files. References *Abramski, K., Citraro, S., Lombardi, L., Rossetti, G., & Stella, M. (2023). Cognitive Network Science Reveals Bias in GPT-3, GPT-3.5 Turbo, and GPT-4 Mirroring Math Anxiety in High-School Students. Big Data and Cognitive Computing , 7 (3), 124. https://doi.org/10.3390/bdcc7030124 Aloisi, C. (2023). The future of standardised assessment: Validity and trust in algorithms for assessment and scoring. European Journal of Education , 58 (1), 98–110. https://doi.org/10.1111/ejed.12542 Amos, J. P., Ahmed Amodu, O., Azlina Raja Mahmood, R., Bolakale Abdulqudus, A., Zakaria, A. F., Rhoda Iyanda, A., Ali Bukar, U., & Mohd Hanapi, Z. (2025). A Bibliometric Exposition and Review on Leveraging LLMs for Programming Education. IEEE Access , 13 , 58364–58393. https://doi.org/10.1109/ACCESS.2025.3554627 *Anand, A., Addala, K., Baghel, K., Goel, A., Hira, M., Gupta, R., & Shah, R. R. (2023). Revolutionizing High School Physics Education: A Novel Dataset. In V. Goyal, N. Kumar, S. S. Bhowmick, P. Goyal, N. Goyal, & D. Kumar (Eds.), Big Data and Artificial Intelligence (Vol. 14418, pp. 64–79). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-49601-1_5 *Anghel, E., Wang, Y., Gopalakrishnan, M., Mansukhani, P., & Bergner, Y. (2024). Can LLMs evaluate items measuring collaborative problem-solving? https://doi.org/10.35542/osf.io/tuvks *Barno, E., Albaladejo-González, M., & Reich, J. (2024). Scaling Generated Feedback for Novice Teachers by Sustaining Teacher Educators’ Expertise: A Design to Train LLMs with Teacher Educator Endorsement of Generated Feedback. Proceedings of the Eleventh ACM Conference on Learning @ Scale , 412–416. https://doi.org/10.1145/3657604.3664677 *Baytak, A. (2024). The Content Analysis of the Lesson Plans Created by ChatGPT and Google Gemini. Research in Social Sciences and Technology , 9 (1), 329–350. https://doi.org/10.46303/ressat.2024.19 *Bellettini, C., Lodi, M., Lonati, V., Monga, M., & Morpurgo, A. (2023). Davinci Goes to Bebras: A Study on the Problem Solving Ability of GPT-3: Proceedings of the 15th International Conference on Computer Supported Education , 59–69. https://doi.org/10.5220/0012007500003470 *Bewersdorff, A., Seßler, K., Baur, A., Kasneci, E., & Nerdel, C. (2023). Assessing student errors in experimentation using artificial intelligence and large language models: A comparative study with human raters. Computers and Education: Artificial Intelligence , 5 , 100177. https://doi.org/10.1016/j.caeai.2023.100177 Bulathwela, S., Pérez-Ortiz, M., Holloway, C., Cukurova, M., & Shawe-Taylor, J. (2024). Artificial Intelligence Alone Will Not Democratise Education: On Educational Inequality, Techno-Solutionism and Inclusive Tools. Sustainability , 16 (2), 781. https://doi.org/10.3390/su16020781 *Butgereit, L., & Van Staden, A. (2023). An AI Contribution to African Economic Development Through Mother Tongue Education—A Linguistic Evaluation of an African-Based AI Mathematics Tutoring Bot. 2023 International Conference on Information and Communication Technology for Development for Africa (ICT4DA) , 7–12. https://doi.org/10.1109/ICT4DA59526.2023.10302241 *Canonigo, A. M. (2024). Levering AI to enhance students’ conceptual understanding and confidence in mathematics. Journal of Computer Assisted Learning , 40 (6), 3215–3229. https://doi.org/10.1111/jcal.13065 *Chen, C.-H., & Chang, C.-L. (2024). Effectiveness of AI-assisted game-based learning on science learning outcomes, intrinsic motivation, cognitive load, and learning behavior. Education and Information Technologies , 29 (14), 18621–18642. https://doi.org/10.1007/s10639-024-12553-x Chen, X., Hu, Z., & Wang, C. (2024). Empowering education development through AIGC: A systematic literature review. Education and Information Technologies , 29 (13), 17485–17537. https://doi.org/10.1007/s10639-024-12549-7 *Cheung, K. K. C., Pun, J. K. H., & Li, W. (2024). Students’ Holistic Reading of Socio-Scientific Texts on Climate Change in a ChatGPT Scenario. Research in Science Education , 54 (5), 957–976. https://doi.org/10.1007/s11165-024-10177-2 Chu, R., & Lim, S. C. J. (2023). Education and Training for Future Engineering Teachers in the Age of Artificial Intelligence: A Bibliometric Analysis. 2023 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM) , 416–420. https://doi.org/10.1109/IEEM58616.2023.10406630 *Cohn, C., Snyder, C., Montenegro, J., & Biswas, G. (2024). Towards A Human-in-the-Loop LLM Approach to Collaborative Discourse Analysis. Artificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky , 2151 , 11–19. https://doi.org/10.1007/978-3-031-64312-5_2 *Cooper, G., & Tang, K.-S. (2024). Pixels and Pedagogy: Examining Science Education Imagery by Generative Artificial Intelligence. Journal of Science Education and Technology , 33 (4), 556–568. https://doi.org/10.1007/s10956-024-10104-0 Curren, R., Boniwell, I., Ryan, R. M., Oades, L., Brighouse, H., Unterhalter, E., Kristjánsson, K., De Ruyter, D., Macleod, C., Morris, I., & White, M. (2024). Finding consensus on well-being in education. Theory and Research in Education , 22 (2), 117–157. https://doi.org/10.1177/14778785241259852 *Daher, W., & Gierdien, F. (2024). Use of Language By generative AI Tools in Mathematical Problem Solving: The Case of ChatGPT. African Journal of Research in Mathematics, Science and Technology Education , 28 (2), 222–235. https://doi.org/10.1080/18117295.2024.2384676 *Dao, X.-Q., & Le, N.-B. (2023). Investigating the Effectiveness of ChatGPT in Mathematical Reasoning and Problem Solving: Evidence from the Vietnamese National High School Graduation Examination (Version 3). arXiv. https://doi.org/10.48550/ARXIV.2306.06331 Darvishi, A., Khosravi, H., Sadiq, S., Gašević, D., & Siemens, G. (2024). Impact of AI assistance on student agency. Computers & Education , 210 , 104967. https://doi.org/10.1016/j.compedu.2023.104967 *Dasari, D., Hendriyanto, A., Sahara, S., Suryadi, D., Muhaimin, L. H., Chao, T., & Fitriana, L. (2024). ChatGPT in didactical tetrahedron, does it make an exception? A case study in mathematics teaching and learning. Frontiers in Education , 8 , 1295413. https://doi.org/10.3389/feduc.2023.1295413 de Ruyter, D., Oades, L. G., Waghid, Y., Ehrenfeld, J., Gilead, T., & Singh, N. C. (2022). Education for flourishing and flourishing in education. In A. K. Duraiappah, N. M. V. Atteveldt, G. Borst, S. Bugden, O. Ergas, T. Gilead, L. Gupta, J. Mercier, K. Pugh, S. N.C., & V. E. A. (Eds.), Reimagining Education: The International Science and Evidence based Education Assessment . UNESCO MGIEP. https://doi.org/10.56383/XBXZ7711 Denecke, K., Glauser, R., & Reichenpfader, D. (2023). Assessing the Potential and Risks of AI-Based Tools in Higher Education: Results from an eSurvey and SWOT Analysis. Trends in Higher Education , 2 (4), 667–688. https://doi.org/10.3390/higheredu2040039 *dos Santos, R. P. (2023). Enhancing Chemistry Learning with ChatGPT, Bing Chat, Bard, and Claude as Agents-to-Think-With: A Comparative Case Study (Version 1). arXiv. https://arxiv.org/abs/2311.00709 *Drushlyak, M., Lukashova, T., Sabadosh, Y., Melnikov, I., & Semenikhina, O. (2024). Using ChatGPT for the Development of Critical Thinking in Youth: Example of Inequality Proof. 2024 47th MIPRO ICT and Electronics Convention (MIPRO) , 334–339. https://doi.org/10.1109/MIPRO60963.2024.10569759 *Due, S., Das, S., Andersen, M., López, B. P., Nexø, S. A., & Clemmensen, L. (2024). Evaluation of Large Language Models: STEM education and Gender Stereotypes (arXiv:2406.10133). arXiv. https://doi.org/10.48550/arXiv.2406.10133 *Egara, F. O., & Mosimege, M. (2024). Exploring the Integration of Artificial Intelligence-Based ChatGPT into Mathematics Instruction: Perceptions, Challenges, and Implications for Educators. Education Sciences , 14 (7), 742. https://doi.org/10.3390/educsci14070742 English, L. D. (2017). Advancing Elementary and Middle School STEM Education. International Journal of Science and Mathematics Education , 15 (S1), 5–24. https://doi.org/10.1007/s10763-017-9802-x *Ergene, O., & Ergene, B. C. (2025). AI ChatBots’ solutions to mathematical problems in interactive e-textbooks: Affordances and constraints from the eyes of students and teachers. Education and Information Technologies , 30 (1), 509–545. https://doi.org/10.1007/s10639-024-13121-z Ergras, O., Gilead, T., Eilam, E., Kyza, E. E., Akar, B., Buck, M. F., & Kumar, S. (2022). A curricular framework for flourishing in education. In A. K. Duraiappah, N. M. V. Atteveldt, G. Borst, S. Bugden, O. Ergas, T. Gilead, L. Gupta, J. Mercier, K. Pugh, S. N.C., & V. E. A. (Eds.), Reimagining Education: The International Science and Evidence based Education Assessment . UNESCO MGIEP. https://doi.org/10.56383/TJKT9645 Farrokhnia, M., Banihashem, S. K., Noroozi, O., & Wals, A. (2024). A SWOT analysis of ChatGPT: Implications for educational practice and research. Innovations in Education and Teaching International , 61 (3), 460–474. https://doi.org/10.1080/14703297.2023.2195846 *Feldman-Maggor, Y., Blonder, R., & Alexandron, G. (2025). Perspectives of Generative AI in Chemistry Education Within the TPACK Framework. Journal of Science Education and Technology , 34 (1), 1–12. https://doi.org/10.1007/s10956-024-10147-3 *Feng, W., Lee, J., McNichols, H., Scarlatos, A., Smith, D., Woodhead, S., Ornelas, N., & Lan, A. (2024). Exploring Automated Distractor Generation for Math Multiple-choice Questions via Large Language Models. Findings of the Association for Computational Linguistics: NAACL 2024 , 3067–3082. https://doi.org/10.18653/v1/2024.findings-naacl.193 *Fernandez, N., Scarlatos, A., Feng, W., Woodhead, S., & Lan, A. (2024). DiVERT: Distractor Generation with Variational Errors Represented as Text for Math Multiple-choice Questions (arXiv:2406.19356). arXiv. https://doi.org/10.48550/arXiv.2406.19356 Filippi, S., & Motyl, B. (2024). Large Language Models (LLMs) in Engineering Education: A Systematic Review and Suggestions for Practical Adoption. Information , 15 (6), 345. https://doi.org/10.3390/info15060345 Flook, L., O’Toole, C., Ritter, J., & Schonert-Reichl, K. A. (2022). Human flourishing in schools. In A. K. Duraiappah, N. M. V. Atteveldt, G. Borst, S. Bugden, O. Ergas, T. Gilead, L. Gupta, J. Mercier, K. Pugh, S. N.C., & V. E. A. (Eds.), Reimagining Education: The International Science and Evidence based Education Assessment . UNESCO MGIEP. https://doi.org/10.56383/FOQI1067 Fock, A., Just, J. & Siller, H.-S. (2025). Opportunities and Limitations of ChatGPT in Creating STEM Tasks: A Case Study with Mathematics Pre-service Teachers. CERME14 , Bozen-Bolzano, Italy. Gao, Y., Zhai, X., Li, M., Lee, G., & Liu, X. (2025). A Multimodal Interactive Framework for Science Assessment in the Era of Generative Artificial Intelligence. Journal of Research in Science Teaching , tea.70009. https://doi.org/10.1002/tea.70009 *Garofalo, S. G., & Farenga, S. J. (2025). Science Teacher Perceptions of the State of Knowledge and Education at the Advent of Generative Artificial Intelligence Popularity. Science & Education , 34 (2), 893–912. https://doi.org/10.1007/s11191-024-00534-y *Gattupalli, S., Lee, W., Allessio, D., Crabtree, D., Arroyo, I., & Woolf, B. (2023). Exploring Pre-Service Teachers’ Perceptions of Large Language Models-Generated Hints in Online Mathematics Learning. AIED2023 Empowering Education with LLMs - the Next-Gen Interface and Content Generation . Geiger, V., Gal, I., & Graven, M. (2023). The connections between citizenship education and mathematics education. ZDM – Mathematics Education , 55 (5), 923–940. https://doi.org/10.1007/s11858-023-01521-3 *Ghazali, S. A., Zaki, N., Ali, L., & Harous, S. (2024). Exploring the Potential of ChatGPT as a Substitute Teacher: A Case Study. International Journal of Information and Education Technology, 14(2), 271–278. https://doi.org/10.18178/ijiet.2024.14.2.2048 *Gregorcic, B., & Pendrill, A.-M. (2023). ChatGPT and the frustrated Socrates. Physics Education , 58 (3), 035021. https://doi.org/10.1088/1361-6552/acc299 *Guler, N. K., Dertli, Z. G., Boran, E., & Yildiz, B. (2024). An artificial intelligence application in mathematics education: Evaluating ChatGPT’s academic achievement in a mathematics exam. Pedagogical Research , 9 (2), em0188. https://doi.org/10.29333/pr/14145 *Han, Z. F., Lin, J., Gurung, A., Thomas, D. R., Chen, E., Borchers, C., Gupta, S., & Koedinger, K. R. (2024). Improving Assessment of Tutoring Practices using Retrieval-Augmented Generation (Version 1). arXiv. https://doi.org/10.48550/ARXIV.2402.14594 Harris, H., & Kittur, J. (2024). Generative Artificial Intelligence in Undergraduate Engineering: A Systematic Literature Review. 2024 ASEE Annual Conference & Exposition Proceedings , 47492. https://doi.org/10.18260/1-2--47492 *Hashem, R., Ali, N., El Zein, F., Fidalgo, P., & Abu Khurma, O. (2024). AI to the rescue: Exploring the potential of ChatGPT as a teacher ally for workload relief and burnout prevention. Research and Practice in Technology Enhanced Learning , 19 , 023. https://doi.org/10.58459/rptel.2024.19023 Helms, M. M., & Nixon, J. (2010). Exploring SWOT analysis – where are we now?: A review of academic research from the last decade. Journal of Strategy and Management , 3 (3), 215–251. https://doi.org/10.1108/17554251011064837 *Henkel, O., Horne-Robinson, H., Dyshel, M., Ch, N., Moreau-Pernet, B., & Abood, R. (2024). Learning to Love Edge Cases in Formative Math Assessment: Using the AMMORE Dataset and Chain-of-Thought Prompting to Improve Grading Accuracy (arXiv:2409.17904). arXiv. https://doi.org/10.48550/arXiv.2409.17904 *Henkel, O., Horne-Robinson, H., Kozhakhmetova, N., & Lee, A. (2024). Effective and Scalable Math Support: Experimental Evidence on the Impact of an AI-Math Tutor in Ghana. In A. M. Olney, I.-A. Chounta, Z. Liu, O. C. Santos, & I. I. Bittencourt (Eds.), Artificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky (Vol. 2150, pp. 373–381). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-64315-6_34 Holstein, K., & Aleven, V. (2022). Designing for human–AI complementarity in K-12 education. AI Magazine , 43 (2), 239–248. https://doi.org/10.1002/aaai.12058 *Hu, B., Zheng, L., Zhu, J., Ding, L., Wang, Y., & Gu, X. (2024). Teaching Plan Generation and Evaluation With GPT-4: Unleashing the Potential of LLM in Instructional Design. IEEE Transactions on Learning Technologies , 17 , 1445–1459. https://doi.org/10.1109/TLT.2024.3384765 Humble, N. (2024). Risk management strategy for generative AI in computing education: How to handle the strengths, weaknesses, opportunities, and threats? International Journal of Educational Technology in Higher Education , 21 (1), 61. https://doi.org/10.1186/s41239-024-00494-x *Jia, J., Wang, T., Zhang, Y., & Wang, G. (2024). The comparison of general tips for mathematical problem solving generated by generative AI with those generated by human teachers. Asia Pacific Journal of Education , 44 (1), 8–28. https://doi.org/10.1080/02188791.2023.2286920 *Jin, H., Yoo, M., Park, J., Lee, Y., Wang, X., & Kim, J. (2024). TeachTune: Reviewing Pedagogical Agents Against Diverse Student Profiles with Simulated Students (Version 3). arXiv. https://doi.org/10.48550/ARXIV.2410.04078 *Kakarla, S., Thomas, D., Lin, J., Gupta, S., & Koedinger, K. R. (2024). Using Large Language Models to Assess Tutors’ Performance in Reacting to Students Making Math Errors (arXiv:2401.03238). arXiv. https://doi.org/10.48550/arXiv.2401.03238 Karakuş, N., Gedik, K., & Kazazoğlu, S. (2025). Ethical Decision-Making in Education: A Comparative Study of Teachers and Artificial Intelligence in Ethical Dilemmas. Behavioral Sciences , 15 (4), 469. https://doi.org/10.3390/bs15040469 Kasneci, E., Sessler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., Günnemann, S., Hüllermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., … Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences , 103 , 102274. https://doi.org/10.1016/j.lindif.2023.102274 Kong, S.-C., & Yang, Y. (2024). A Human-Centered Learning and Teaching Framework Using Generative Artificial Intelligence for Self-Regulated Learning Development Through Domain Knowledge Learning in K–12 Settings. IEEE Transactions on Learning Technologies , 17 , 1562–1573. https://doi.org/10.1109/TLT.2024.3392830 Korseberg, L., & Elken, M. (2025). Waiting for the revolution: How higher education institutions initially responded to ChatGPT. Higher Education , 89 (4), 953–968. https://doi.org/10.1007/s10734-024-01256-4 *Küchemann, S., Steinert, S., Revenga, N., Schweinberger, M., Dinc, Y., Avila, K. E., & Kuhn, J. (2023). Can ChatGPT support prospective teachers in physics task development? Physical Review Physics Education Research , 19 (2), 020128. https://doi.org/10.1103/PhysRevPhysEducRes.19.020128 Kuckartz, U. (2019). Qualitative Text Analysis: A Systematic Approach. In G. Kaiser & N. Presmeg (Eds.), Compendium for Early Career Researchers in Mathematics Education (pp. 181–197). Springer International Publishing. https://doi.org/10.1007/978-3-030-15636-7_8 Kuckartz, U., & Rädiker, S. (2023). Qualitative content analysis: Methods, practice and software (2nd edition). SAGE. *Kunz, S., & Steffen, A. (2024). Criticize my Code, not me: Using AI-generated Feedback in Computer Science Teaching. International Conference on Gender Research , 7 (1), 200–207. https://doi.org/10.34190/icgr.7.1.2002 *Latif, E., & Zhai, X. (2024). Fine-tuning ChatGPT for automatic scoring. Computers and Education: Artificial Intelligence , 6 , 100210. https://doi.org/10.1016/j.caeai.2024.100210 *Lee, G.-G., Latif, E., Wu, X., Liu, N., & Zhai, X. (2024). Applying large language models and chain-of-thought for automatic scoring. Computers and Education: Artificial Intelligence , 6 , 100213. https://doi.org/10.1016/j.caeai.2024.100213 *Levonian, Z., Li, C., Zhu, W., Gade, A., Henkel, O., Postle, M.-E., & Xing, W. (2023). Retrieval-augmented Generation to Improve Math Question-Answering: Trade-offs Between Groundedness and Human Preference (arXiv:2310.03184). arXiv. https://doi.org/10.48550/arXiv.2310.03184 *Li, R., Wang, Y., Zheng, C., Jiang, Y.-H., & Jiang, B. (2024). Generating Contextualized Mathematics Multiple-Choice Questions Utilizing Large Language Models. In A. M. Olney, I.-A. Chounta, Z. Liu, O. C. Santos, & I. I. Bittencourt (Eds.), Artificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky (Vol. 2150, pp. 494–501). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-64315-6_48 *Li, T., Ji, Y., & Zhan, Z. (2024). Expert or machine? Comparing the effect of pairing student teacher with in-service teacher and ChatGPT on their critical thinking, learning performance, and cognitive load in an integrated-STEM course. Asia Pacific Journal of Education , 44 (1), 45–60. https://doi.org/10.1080/02188791.2024.2305163 *Li, Z., & Ironsi, C. S. (2024). The Efficacy of Generative Artificial Intelligence in Developing Science Education Preservice Teachers’ Writing Skills: An Experimental Approach. Journal of Science Education and Technology . https://doi.org/10.1007/s10956-024-10148-2 *Li, Z.-Z., Zhang, M.-L., Yin, F., Ji, Z.-L., Bai, J.-F., Pan, Z.-R., Zeng, F.-H., Xu, J., Zhang, J.-X., & Liu, C.-L. (2024). CMMaTH: A Chinese Multi-modal Math Skill Evaluation Benchmark for Foundation Models (arXiv:2407.12023). arXiv. https://doi.org/10.48550/arXiv.2407.12023 *Lin, T., & Riccomini, P. (2024). Inclusive Education in Rural Settings: Leveraging Technology for Improved Math Learning Outcomes Among Students With Disabilities. Journal of Special Education Technology , 40 (2), 219–226. https://doi.org/10.1177/01626434241263045 Listyaningrum, P., Retnawati, H., Harun, H., & Ibda, H. (2024). Digital learning using ChatGPT in elementary school mathematics learning: A systematic literature review. Indonesian Journal of Electrical Engineering and Computer Science , 36 (3), 1701. https://doi.org/10.11591/ijeecs.v36.i3.pp1701-1710 *Liu, N., Sonkar, S., Le, M., & Baraniuk, R. (2024). MalAlgoQA: Pedagogical Evaluation of Counterfactual Reasoning in Large Language Models and Implications for AI in Education (arXiv:2407.00938). arXiv. https://doi.org/10.48550/arXiv.2407.00938 Loke, D., Radloff, J., & Yeter, I. H. (2024). Board 165: K-12 STEM Teachers’ Perceptions of Artificial Intelligence: A PRISMA-tic Approach (Work-in-Progress). 2024 ASEE Annual Conference & Exposition Proceedings , 46727. https://doi.org/10.18260/1-2--46727 *Long, Y., Luo, H., & Zhang, Y. (2024). Evaluating large language models in analysing classroom dialogue. Npj Science of Learning , 9 (1), 60. https://doi.org/10.1038/s41539-024-00273-3 *Lu, J., Zheng, R., Gong, Z., & Xu, H. (2024). Supporting Teachers’ Professional Development With Generative AI: The Effects on Higher Order Thinking and Self-Efficacy. IEEE Transactions on Learning Technologies , 17 , 1267–1277. https://doi.org/10.1109/TLT.2024.3369690 Mayring, P. (2014). Qualitative content analysis: Theoretical foundation, basic procedures and software solution . http://nbn-resolving.de/urn:nbn:de:0168-ssoar-395173 McGowan, J., Sampson, M., Salzwedel, D. M., Cogo, E., Foerster, V., & Lefebvre, C. (2016). PRESS Peer Review of Electronic Search Strategies: 2015 Guideline Statement. Journal of Clinical Epidemiology , 75 , 40–46. https://doi.org/10.1016/j.jclinepi.2016.01.021 *McNichols, H., Feng, W., Lee, J., Scarlatos, A., Smith, D., Woodhead, S., & Lan, A. (2023). Automated Distractor and Feedback Generation for Math Multiple-choice Questions via In-context Learning (arXiv:2308.03234). arXiv. https://doi.org/10.48550/arXiv.2308.03234 *McNichols, H., Lee, J., Fancsali, S., Ritter, S., & Lan, A. (2024). Can Large Language Models Replicate ITS Feedback on Open-Ended Math Questions? (arXiv:2405.06414). arXiv. https://doi.org/10.48550/arXiv.2405.06414 Merzifonluoglu, A., & Gunes, H. (2025). Shifting Dynamics: Who Holds the Reins in Decision-Making With Artificial Intelligence Tools? Perspectives of Gen Z Pre‐Service Teachers. European Journal of Education , 60 (1), e70053. https://doi.org/10.1111/ejed.70053 Miao, F., & Cukurova, M. (2024). AI competency framework for teachers . UNESCO. https://doi.org/10.54675/ZJTE2084 Miao, F., & Holmes, W. (2023). Guidance for generative AI in education and research . UNESCO. https://doi.org/10.54675/EWZM9535 Mishra, P., Warr, M., & Islam, R. (2023). TPACK in the age of ChatGPT and Generative AI. Journal of Digital Learning in Teacher Education , 39 (4), 235–251. https://doi.org/10.1080/21532974.2023.2247480 *Moreau-Pernet, B., Tian, Y., Sawaya, S., Foltz, P., Cao, J., Milne, B., & Christie, T. (2024). Classifying Tutor Discursive Moves at Scale in Mathematics Classrooms with Large Language Models. Proceedings of the Eleventh ACM Conference on Learning @ Scale , 361–365. https://doi.org/10.1145/3657604.3664664 Munn, Z., Peters, M. D. J., Stern, C., Tufanaru, C., McArthur, A., & Aromataris, E. (2018). Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach. BMC Medical Research Methodology , 18 (1), 143. https://doi.org/10.1186/s12874-018-0611-x Navigli, R., Conia, S., & Ross, B. (2023). Biases in Large Language Models: Origins, Inventory, and Discussion. Journal of Data and Information Quality , 15 (2), 1–21. https://doi.org/10.1145/3597307 Ng, D. T. K., Leung, J. K. L., Su, J., Ng, R. C. W., & Chu, S. K. W. (2023). Teachers’ AI digital competencies and twenty-first century skills in the post-pandemic world. Educational Technology Research and Development , 71 (1), 137–161. https://doi.org/10.1007/s11423-023-10203-6 *Nicula, B., Dascalu, M., Arner, T., Balyan, R., & McNamara, D. S. (2023). Automated Assessment of Comprehension Strategies from Self-Explanations Using LLMs. Information , 14 (10), 567. https://doi.org/10.3390/info14100567 *Norberg, K. A., Almoubayyed, H., De Ley, L., Murphy, A., Weldon, K., & Ritter, S. (2024). Rewriting Content with GPT-4 to Support Emerging Readers in Adaptive Mathematics Software. International Journal of Artificial Intelligence in Education . https://doi.org/10.1007/s40593-024-00420-2 OECD. (2023). The New Flourishing Agenda in Education: A Report on the Current Theoretical State of Play . https://www.oecd.org/content/dam/oecd/en/about/projects/edu/hpst/EHF_The%20new%20flourishing%20agenda_Kristjansson.pdf OECD. (2024a). PISA 2022 Results (Volume V): Learning Strategies and Attitudes for Life . OECD. https://doi.org/10.1787/c2e44201-en OECD. (2024b). PISA High Performing Systems for Tomorrow: Education for Human Flourishing . https://one.oecd.org/document/EDU/PISA/GB(2024)10/en/pdf Opesemowo, O. A. G., & Adewuyi, H. O. (2024). A systematic review of artificial intelligence in mathematics education: The emergence of 4IR. Eurasia Journal of Mathematics, Science and Technology Education , 20 (7), em2478. https://doi.org/10.29333/ejmste/14762 *Parra, V., Sureda, P., Corica, A., Schiaffino, S., & Godoy, D. (2024). Can Generative AI Solve Geometry Problems? Strengths and Weaknesses of LLMs for Geometric Reasoning in Spanish. International Journal of Interactive Multimedia and Artificial Intelligence , 8 (5), 65. https://doi.org/10.9781/ijimai.2024.02.009 *Pavlova, N. H. (2024). Flipped dialogic learning method with ChatGPT: A case study. International Electronic Journal of Mathematics Education , 19 (1), em0764. https://doi.org/10.29333/iejme/14025 Pendleton, D. (1975). Calculators in the classroom. Science News , 107 , 175–181. Pepin, B., Buchholtz, N., & Salinas-Hernández, U. (2025). A Scoping Survey of ChatGPT in Mathematics Education. Digital Experiences in Mathematics Education , 11 (1), 9–41. https://doi.org/10.1007/s40751-025-00172-1 *Prihar, E., Lee, M., Hopman, M., Kalai, A. T., Vempala, S., Wang, A., Wickline, G., Murray, A., & Heffernan, N. (2023). Comparing Different Approaches to Generating Mathematics Explanations Using Large Language Models. In N. Wang, G. Rebolledo-Mendez, V. Dimitrova, N. Matsuda, & O. C. Santos (Eds.), Artificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky (Vol. 1831, pp. 290–295). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-36336-8_45 Raihan, N., Siddiq, M. L., Santos, J. C. S., & Zampieri, M. (2024). Large Language Models in Computer Science Education: A Systematic Literature Review (arXiv:2410.16349). arXiv. https://doi.org/10.48550/arXiv.2410.16349 Rethlefsen, M. L., Kirtley, S., Waffenschmidt, S., Ayala, A. P., Moher, D., Page, M. J., Koffel, J. B., PRISMA-S Group, Blunt, H., Brigham, T., Chang, S., Clark, J., Conway, A., Couban, R., De Kock, S., Farrah, K., Fehrmann, P., Foster, M., Fowler, S. A., … Young, S. (2021). PRISMA-S: An extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews. Systematic Reviews , 10 (1), 39. https://doi.org/10.1186/s13643-020-01542-z *Ribeiro, A. R. A., Rosotti Navarro, E., & Kalinke, M. A. (2024). O uso do ChatGPT para resolver problemas matemáticos sobre grandezas direta e inversamente proporcionais. Revista Pesquisa Qualitativa , 12 (30), 01–21. https://doi.org/10.33361/RPQ.2024.v.12.n.30.716 Roberts, J., & Mohamed, A. (2024). Generative AI in CS Education: Literature Review through a SWOT Lens. The 26th Western Canadian Conference on Computing Education , 1–6. https://doi.org/10.1145/3660650.3660657 *Rouzegar, H., & Makrehchi, M. (2024). Generative AI for Enhancing Active Learning in Education: A Comparative Study of GPT-3.5 and GPT-4 in Crafting Customized Test Questions . https://doi.org/10.48550/ARXIV.2406.13903 Russell, S., & Norvig, P. (2021). Artificial Intelligence: A Modern Approach (4th ed.). Pearson. https://elibrary.pearson.de/book/99.150005/9781292401171 Schinkel, A., Wolbert, L., Pedersen, J. B. W., & De Ruyter, D. J. (2023). Human Flourishing, Wonder, and Education. Studies in Philosophy and Education , 42 (2), 143–162. https://doi.org/10.1007/s11217-022-09851-7 *Schorcht, S., Baumanns, L., Buchholtz, N., Huget, J., Peters, F., & Pohl, M. (2024). The tasty world of irrational numbers – explorative mathematical visualizations generated by AI image models. Journal of Mathematics and the Arts , 18 (3–4), 272–289. https://doi.org/10.1080/17513472.2024.2407273 *Schorcht, S., Buchholtz, N., & Baumanns, L. (2024). Prompt the problem – investigating the mathematics educational quality of AI-supported problem solving by comparing prompt techniques. Frontiers in Education , 9 , 1386075. https://doi.org/10.3389/feduc.2024.1386075 *Shankar, S. K., Pothancheri, G., Sasi, D., & Mishra, S. (2025). Bringing Teachers in the Loop: Exploring Perspectives on Integrating Generative AI in Technology-Enhanced Learning. International Journal of Artificial Intelligence in Education , 35 (1), 155–180. https://doi.org/10.1007/s40593-024-00428-8 Siller, H.-S., Fock, A., & Just, J. (2025). Mathematics, Sustainability and Artificial Intelligence – Reflecting the Challenges, Potentials & Opportunities of the Socio-ecological in Mathematics Lessons . Siller, H.-S., Günster, S. M., & Geiger, V. (2024). Mathematics as a Central Focus in STEM – Theoretical and Practical Insights from a Special Study Program Within Pre-service (Prospective) Teacher Education. In Y. Li, Z. Zeng, & N. Song (Eds.), Disciplinary and Interdisciplinary Education in STEM (pp. 317–343). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-52924-5_15 *Singh, A., Ehtesham, A., Kumar, S., Gupta, G. K., & Khoei, T. T. (2024). Encouraging Responsible Use of Generative AI in Education: A Reward-Based Learning Approach. https://arxiv.org/abs/2407.15022v1 * Sonkar, S., Chen, X., Le, M., Liu, N [Naiming], Basu Mallick, D., & Baraniuk, R [Richard] (2024). Code Soliloquies for Accurate Calculations in Large Language Models. In Proceedings of the 14th Learning Analytics and Knowledge Conference (pp. 828–835). ACM. https://doi.org/10.1145/3636555.3636889 *Spreitzer, C., Straser, O., Zehetmeier, S., & Maaß, K. (2024). Mathematical Modelling Abilities of Artificial Intelligence Tools: The Case of ChatGPT. Education Sciences , 14 (7), 698. https://doi.org/10.3390/educsci14070698 Stevenson, M. (2022). Education for Human Flourishing . Su, F., & Jackson, C. (2020). Mathematics for human flourishing . Yale University Press. *Taani, O., & Alabidi, S. (2024). ChatGPT in education: Benefits and challenges of ChatGPT for mathematics and science teaching practices. International Journal of Mathematical Education in Science and Technology , 1–30. https://doi.org/10.1080/0020739X.2024.2357341 Taulli, T. (2023). Generative AI: How ChatGPT and Other AI Tools Will Revolutionize Business . Apress. https://doi.org/10.1007/978-1-4842-9367-6 *Teegavarapu, R. R., & Sanghvi, H. (2023). Analyzing the Competitive Mathematical Problem-Solving Skills of ChatGPT. 2023 International Conference on Data Science, Agents & Artificial Intelligence (ICDSAAI) , 1–7. https://doi.org/10.1109/ICDSAAI59313.2023.10452659 Tricco, A. C., Lillie, E., Zarin, W., O’Brien, K. K., Colquhoun, H., Levac, D., Moher, D., Peters, M. D. J., Horsley, T., Weeks, L., Hempel, S., Akl, E. A., Chang, C., McGowan, J., Stewart, L., Hartling, L., Aldcroft, A., Wilson, M. G., Garritty, C., … Straus, S. E. (2018). PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Annals of Internal Medicine , 169 (7), 467–473. https://doi.org/10.7326/M18-0850 *van Pham, P. L., Duc, A. V., Hoang, N. M., Do, X. L., & Luu, A. T. (2024). ChatGPT as a Math Questioner? Evaluating ChatGPT on Generating Pre-university Math Questions. In J. Hong, J. W. Park, & A. Przybyłek (Eds.), Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing (pp. 65–73). ACM. https://doi.org/10.1145/3605098.3636030 *Vankúš, P. (2024). Generative Artificial Intelligence on Mobile Devices in the University Preparation of Future Teachers of Mathematics. International Journal of Interactive Mobile Technologies (iJIM) , 18 (18), 19–33. https://doi.org/10.3991/ijim.v18i18.51221 *Vasconcelos, M. A. R., & Dos Santos, R. P. (2023). Enhancing STEM learning with ChatGPT and Bing Chat as objects to think with: A case study. Eurasia Journal of Mathematics, Science and Technology Education , 19 (7), em2296. https://doi.org/10.29333/ejmste/13313 Virvou, M., & Tsihrintzis, G. A. (2024). Artificial Intelligence Chatbots and Social Robots in Education: FEPER Framework for Efficiency, Pedagogical and Ethical Requirements. 2024 15th International Conference on Information, Intelligence, Systems & Applications (IISA) , 1–8. https://doi.org/10.1109/IISA62523.2024.10786710 Von Davier, M., Kennedy, A., Reynolds, K., Fishbein, B., Khorramdel, L., Aldrich, C., Bookbinder, A., Bezirhan, U., & Yin, L. (2024). TIMSS 2023 International Results in Mathematics and Science . TIMSS & PIRLS International Study Center, Boston College. https://doi.org/10.6017/lse.tpisc.timss.rs6460 *Wardat, Y., Tashtoush, M. A., AlAli, R., & Jarrah, A. M. (2023). ChatGPT: A revolutionary tool for teaching and learning mathematics. Eurasia Journal of Mathematics, Science and Technology Education , 19 (7), em2286. https://doi.org/10.29333/ejmste/13272 *Wei, X. (2024). Evaluating chatGPT-4 and chatGPT-4o: Performance insights from NAEP mathematics problem solving. Frontiers in Education , 9 , 1452570. https://doi.org/10.3389/feduc.2024.1452570 Willoughby, S. (1985). Curriculum trends: Mathematics: Will calculators rot our minds? Educational Leadership , 43 , 90–91. Wolbert, L. S., De Ruyter, D. J., & Schinkel, A. (2015). Formal criteria for the concept of human flourishing: The first step in defending flourishing as an ideal aim of education. Ethics and Education , 10 (1), 118–129. https://doi.org/10.1080/17449642.2014.998032 Xu, W., & Ouyang, F. (2022). A systematic review of AI role in the educational system based on a proposed conceptual framework. Education and Information Technologies , 27 (3), 4195–4223. https://doi.org/10.1007/s10639-021-10774-y Yacoubian, H. A. (2018). Scientific literacy for democratic decision-making. International Journal of Science Education , 40 (3), 308–327. https://doi.org/10.1080/09500693.2017.1420266 *Yan, Y., Wang, S., Huo, J., Li, H., Li, B., Su, J., Gao, X., Zhang, Y.-F., Xu, T., Chu, Z., Zhong, A., Wang, K., Xiong, H., Yu, P. S., Hu, X., & Wen, Q. (2024). ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection (Version 2). arXiv. https://doi.org/10.48550/ARXIV.2410.04509 Yang, Y., Chen, L., He, W., Sun, D., & Salas-Pilco, S. Z. (2024). Artificial Intelligence for Enhancing Special Education for K-12: A Decade of Trends, Themes, and Global Insights (2013–2023). International Journal of Artificial Intelligence in Education . https://doi.org/10.1007/s40593-024-00422-0 *Yeadon, W., & Hardy, T. (2024). The impact of AI in physics education: A comprehensive review from GCSE to university levels. Physics Education , 59 (2), 025010. https://doi.org/10.1088/1361-6552/ad1fa2 *Yue, M., Lyu, W., Mifdal, W., Suh, J., Zhang, Y., & Yao, Z. (2024). MathVC: An LLM-Simulated Multi-Character Virtual Classroom for Mathematics Education (Version 2). arXiv. https://doi.org/10.48550/ARXIV.2404.06711 *Yunianto, W., Lavicza, Z., Kastner-Hauler, O., & Houghton, T. (2024). Investigating the use of ChatGPT to solve a GeoGebra based mathematics + computational thinking task in a geometry topic. Journal on Mathematics Education , 15 (3), 1027–1052. https://doi.org/10.22342/jme.v15i3.pp1027-1052 Yusuf, A., Pervin, N., Román-González, M., & Noor, N. M. (2024). Generative AI in education and research: A systematic mapping review. Review of Education , 12 (2), e3489. https://doi.org/10.1002/rev3.3489 *Zhang, F., Li, C., Henkel, O., Xing, W., Baral, S., Heffernan, N., & Li, H. (2024). Math-LLMs: AI Cyberinfrastructure with Pre-trained Transformers for Math Education. International Journal of Artificial Intelligence in Education . https://doi.org/10.1007/s40593-024-00416-y Footnotes By “secondary STEM education” we mean both education in the individual STEM disciplines (e.g. mathematics education) and the combination or integration of several STEM disciplines within an educational setting. We refrain from an in-depth discussion of the relationship and integration of the various disciplines (see English, 2017 , for an overview). We decided not to include technology directly as a discipline to avoid irrelevant results, which are not related to technology education but used to term technology , for example to state that ChatGPT is an emerging technology. We suppose that this restriction does not exclude relevant results. Retrieval augmented generation (RAG) is a possibility make a generative AI system use an external data base to retrieve information. Additional Declarations No competing interests reported. Supplementary Files Listofincludedpublications.pdf Codingscheme.pdf Cite Share Download PDF Status: Published Journal Publication published 01 Dec, 2025 Read the published version in International Journal of STEM Education → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6923010","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":477804354,"identity":"0b82a143-8565-480b-a0bd-8c448d056e61","order_by":0,"name":"Alissa Fock","email":"","orcid":"","institution":"University of Würzburg","correspondingAuthor":false,"prefix":"","firstName":"Alissa","middleName":"","lastName":"Fock","suffix":""},{"id":477804357,"identity":"7ac3ff8f-20f4-4c62-a0b0-3f9f41d58fd5","order_by":1,"name":"Hans-Stefan Siller","email":"data:image/png;base64,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","orcid":"","institution":"University of Würzburg","correspondingAuthor":true,"prefix":"","firstName":"Hans-Stefan","middleName":"","lastName":"Siller","suffix":""}],"badges":[],"createdAt":"2025-06-18 12:23:22","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6923010/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6923010/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s40594-025-00589-5","type":"published","date":"2025-12-01T15:57:38+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":85824194,"identity":"a9f9d571-5628-4a36-a66d-f72f2c60052e","added_by":"auto","created_at":"2025-07-02 06:58:54","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":72383,"visible":true,"origin":"","legend":"\u003cp\u003eFlow Diagram of the Scoping Review\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/25f33b8ace075d1635b7e517.png"},{"id":85822806,"identity":"9543466e-41f3-42d1-8372-37dad7024aa9","added_by":"auto","created_at":"2025-07-02 06:50:54","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":31756,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eDisciplines covered by the publications\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/2e7440815751c7efb6bc8dc5.png"},{"id":85825123,"identity":"be32b62d-497c-4d3f-9388-2824e38ba27c","added_by":"auto","created_at":"2025-07-02 07:06:54","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":66580,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eDevelopment of publications over time\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/ebf4e7c15b95d72ce864d6d3.png"},{"id":85824197,"identity":"d66cfafd-4c6d-4ba0-9684-4732eb538abf","added_by":"auto","created_at":"2025-07-02 06:58:54","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":42408,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eTypes of journals and conferences\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/722edca1109c53ef14aa44a9.png"},{"id":85822814,"identity":"b0d3e6bb-50ed-4b8e-915f-07cceb3b206e","added_by":"auto","created_at":"2025-07-02 06:50:54","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":41834,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eTypes of publications and research methods\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/cbdc6163e29a417b03ca2107.png"},{"id":85825124,"identity":"471e3e19-17bb-4f61-a103-6d09504fa136","added_by":"auto","created_at":"2025-07-02 07:06:54","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":34025,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eData sources; light shade signifies that the data source is used in conjunction with another data source\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/4912b70985693e797556867b.png"},{"id":85824200,"identity":"59130235-b484-4fc4-8e1f-9a5171de3211","added_by":"auto","created_at":"2025-07-02 06:58:55","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":68746,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eResearch themes overall (black) and divided by discipline (colored); bars are higher for disciplines because publications addressing multiple disciplines are counted multiple times for the respective research theme\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image7.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/ddb391ac92c4c86907b1d5a8.png"},{"id":85822818,"identity":"76930612-ecbd-4471-a11c-185d36f96e5f","added_by":"auto","created_at":"2025-07-02 06:50:55","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":123559,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eSubcategorization of the research themes of performance and tool \u0026amp; dataset development; \u003c/em\u003e\u003cu\u003eidentical colors\u003c/u\u003e\u003cem\u003e refer to the same subcategory\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image8.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/04fa1e9bc1fc081b4812904e.png"},{"id":85822819,"identity":"0064d132-dc30-4367-87e3-4e2fea4b0fcf","added_by":"auto","created_at":"2025-07-02 06:50:55","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":48153,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eDevelopment over time and participants split by research theme\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image9.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/9d8df528d5258d38cb3dbb22.png"},{"id":85822863,"identity":"1272bb6f-ab5b-4505-98ec-2e65ccc81462","added_by":"auto","created_at":"2025-07-02 06:50:56","extension":"jpeg","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":501546,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eSWOT analysis of generative AI in secondary STEM education, including strengths (upper left), weaknesses (upper right), opportunities (lower left) and threats (lower right), for students (light blue/red), teachers (medium blue/red) and pre-service teachers (dark blue/red)\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image10.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/5de900043a2bb4a856929af7.jpeg"},{"id":85824206,"identity":"0acf22c8-132f-4327-8329-28047c3a91c9","added_by":"auto","created_at":"2025-07-02 06:58:56","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":74891,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eStrategy for teacher-AI-collaboration based on SWOT analysis\u003c/em\u003e\u003c/p\u003e","description":"","filename":"image11.png","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/1f778d31fab1b30d8a513fd4.png"},{"id":97724811,"identity":"91c66086-aeec-4c25-b92a-fe308ea1f421","added_by":"auto","created_at":"2025-12-08 16:13:35","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2349563,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/4b073120-0e52-4f66-a26e-af1110c2cb20.pdf"},{"id":85822808,"identity":"fcb22823-5751-4f48-b943-351f6e2921e9","added_by":"auto","created_at":"2025-07-02 06:50:54","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":466893,"visible":true,"origin":"","legend":"","description":"","filename":"Listofincludedpublications.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/ef9c4ed080f963296a050cfe.pdf"},{"id":85822810,"identity":"3b9430aa-62fb-4e88-9d16-2504738f1918","added_by":"auto","created_at":"2025-07-02 06:50:54","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":239106,"visible":true,"origin":"","legend":"","description":"","filename":"Codingscheme.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6923010/v1/ba98e061d2a63e6f7447e698.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Generative Artificial Intelligence in Secondary STEM Education in the Light of Human Flourishing: A Scoping Literature Review","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe adoption of emerging technologies in STEM education has long been double-edged. With each new tool \u0026ndash; be it the pocket calculator or the rise of computer algebra systems \u0026ndash; comes the debate of its promises and perils in the classroom. Regarding pocket calculators, which eased calculations in various STEM disciplines, concerns have been voiced already in the 1980s that these tools \u0026ldquo;will discourage students from using their minds\u0026rdquo; and that school will \u0026ldquo;graduate people who are unable to think quantitatively and who depend on calculating devices\u0026rdquo; (Willoughby, \u003cspan citationid=\"CR131\" class=\"CitationRef\"\u003e1985\u003c/span\u003e, p. 90). These fears of overdependency and loss of fundamental skills stood opposed to great expectations of calculators allowing students to tackle more complex and realistic problems without being constrained by long computations (Pendleton, \u003cspan citationid=\"CR100\" class=\"CitationRef\"\u003e1975\u003c/span\u003e). With advancements in Artificial Intelligence (AI) and its potential integration into STEM education, arguments and debates surrounding earlier technologies appear to resurface.\u003c/p\u003e \u003cp\u003eThe term \u0026ldquo;Artificial Intelligence\u0026rdquo; is notoriously hard to define (see, e.g., Russell \u0026amp; Norvig, \u003cspan citationid=\"CR108\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). A subset of AI, which has seen a massive surge in public attention since the release of ChatGPT in late 2022, is \u003cem\u003eGenerative AI. Generative AI\u003c/em\u003e refers to AI systems that are able to create new content (such as text, images or videos) based on their training data (Taulli, \u003cspan citationid=\"CR121\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Following the current discourse on generative AI \u0026ndash; particularly ChatGPT \u0026ndash; in the classroom, we notice striking similarities between statements made today and those on former educational technologies. Still, educational institutions perceive that the emergence of generative AI \u0026ldquo;\u003cem\u003efeels\u003c/em\u003e different\u0026rdquo; (Korseberg \u0026amp; Elken, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2025\u003c/span\u003e, p. 960). Developments regarding generative AI fundamentally shake the role of human beings in the teaching and learning process. Probably more than previous educational technologies they impact human decision-making, identity and the purpose of being human \u0026ndash; in other words, they affect \u003cem\u003eHuman Flourishing\u003c/em\u003e, the ability of humans to lead a meaningful, autonomous life by developing their full potential. To ensure that the adoption of generative AI in education does not impede but foster the flourishing of students and teachers alike, its effective, ethical and evidence-based implementation is of pivotal importance \u0026ndash; throughout different disciplines and educational levels.\u003c/p\u003e \u003cp\u003eThis scoping literature review provides an overview of the current research on generative AI in secondary STEM education\u003csup\u003e1\u003c/sup\u003e, focusing particularly on research themes addressed in the literature and on the strengths, weaknesses, opportunities and threats of generative AI in secondary STEM education emerging from empirical evidence. Based on the findings, it is investigated to what extent the current research provides a foundation for exploring Human Flourishing and AI in secondary STEM education.\u003c/p\u003e"},{"header":"2. Theoretical Background","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 The Notion of \u003cem\u003eHuman Flourishing\u003c/em\u003e in Education\u003c/h2\u003e \u003cp\u003e \u003cem\u003eHuman Flourishing\u003c/em\u003e has been identified by various sources as one central goal of education (Curren et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; OECD, \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e2023\u003c/span\u003e, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e; Schinkel et al., \u003cspan citationid=\"CR109\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The exact definition of the term depends on philosophical, psychological, political, academic and cultural backgrounds (Curren et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; de Ruyter et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Broadly, the term \u003cem\u003eHuman Flourishing\u003c/em\u003e refers to the ongoing realization of a person\u0026rsquo;s fullest potential and living well as a human, which includes engagement in meaningful relationships and activities (de Ruyter et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Wolbert et al., \u003cspan citationid=\"CR132\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). A person leading a flourishing life is able to act and decide freely and independently (a capacity referred to as \u003cem\u003eagency\u003c/em\u003e) (de Ruyter et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Stevenson, \u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and \u0026ldquo;has found meaning and purpose through autonomous decision-making\u0026rdquo; (OECD, \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e2023\u003c/span\u003e, p. 22). Flourishing people contribute not only to their own flourishing but to the flourishing of other people in society (Curren et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Non-Western accounts of human flourishing particularly emphasize this notion that individual flourishing is deeply relational and fundamentally connected to the flourishing of one\u0026rsquo;s community (OECD, \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe aim of education for Human Flourishing is to \u0026ldquo;increase [students\u0026rsquo;] ability to choose and follow their own path towards a flourishing life\u0026rdquo; and actively engage in creating and upholding circumstances that allow flourishing for everyone (Ergras et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), considering both students\u0026rsquo; present and future lives (Schinkel et al., \u003cspan citationid=\"CR109\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). This requires a holistic view of the student, extending beyond cognitive capabilities to socio-emotional and personal capacities, targeting students\u0026rsquo; critical reflection, ethical decision making and perspective-taking (Flook et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Human Flourishing consists of intertwining scientific and moral reasoning, fostering both adaptive problem solving \u0026ndash; including creative and critical thinking, synthesis skills and collaboration \u0026ndash; and ethical decision making, i.e., balancing the needs of various individuals and making fair, altruistic choices (Stevenson, \u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Successful education considers that students differ in their potential, wishes and thoughts about a desirable life, and recognizes different, individual ways of flourishing. While education should nurture these capabilities to enable Human Flourishing, flourishing students and teachers also enhance the quality of education (de Ruyter et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This implies that not only students\u0026rsquo; but also teachers\u0026rsquo; flourishing needs to be considered (Curren et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Human Flourishing and AI in Education\u003c/h2\u003e \u003cp\u003eAs students\u0026rsquo; present and future lives are and will increasingly be infused by technology, technology plays a crucial role in students\u0026rsquo; flourishing, enabling them to be responsible citizens in a digitized society (Ergras et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). While this holds true for a variety of technologies, the recent spark of attention in AI calls for a more detailed view of this particular technological development and its implications for Human Flourishing. To this end, the OECD (\u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e) has put forward three themes: \u003cem\u003eAI and broadening human capabilities\u003c/em\u003e, \u003cem\u003eAI and developing new models for the future\u003c/em\u003e, and \u003cem\u003eAI and restoring meaning to individual lives\u003c/em\u003e. We take these three themes as a starting point for elaborating on the interaction between Human Flourishing and AI in education.\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.2.1 Human Flourishing Education and AI: Broadening Human Capabilities\u003c/h2\u003e \u003cp\u003eState-of-the-art (generative) AI systems are capable of tackling tasks that were previously reserved for humans, such as producing natural language and solving mathematical problems. These recent developments of AI underline the importance of fostering students\u0026rsquo; socio-cognitive and meta-cognitive rather than purely cognitive skills, cultivating a holistic view that is distinct for human intelligence (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e; Stevenson, \u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). They call for a particular emphasis on ethical decision-making; while AI systems make analytic, solution-oriented choices based on algorithms, humans are able to make altruistic and empathetic decisions (Karakuş et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Stevenson, \u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), taking into account \u0026ldquo;complex and emotional elements such as values, ethical principles, and moral standards\u0026rdquo; (Merzifonluoglu \u0026amp; Gunes, \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2025\u003c/span\u003e, p. 12).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.2.2 Human Flourishing Education and AI: Developing new Models for the Future\u003c/h2\u003e \u003cp\u003eWith the advent of advanced AI systems, some established models in politics, economics and society are increasingly prone to fraud. Thus, on the one hand, AI systems urge us to develop new societal, economic and organizational models and on the other hand, they can support the development of new models and structures in order to enable Human Flourishing (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e). Within the realm of education, increasingly human-like AI systems are challenging human agency and accountability in teaching and learning (Darvishi et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Merzifonluoglu \u0026amp; Gunes, \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) and are reshaping teacher-student interactions (Xu \u0026amp; Ouyang, \u003cspan citationid=\"CR133\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Concerns are raised about generative AI systems endangering the integrity in exams and homework (Gao et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), thereby undermining established educational practices. These changes suggest the need for new models within the educational landscape (Gao et al., \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Xu \u0026amp; Ouyang, \u003cspan citationid=\"CR133\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Teachers are primary actors in educational processes, mediating their students\u0026rsquo; use of AI (Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e); there is thus a particular need to reconsider teachers\u0026rsquo; roles and competencies in the light of (generative) AI technologies (Feldman-Maggor et al., 2025; Fock et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Mishra et al., \u003cspan citationid=\"CR87\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Ng et al., \u003cspan citationid=\"CR91\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Pivotal to these undertakings is that new models and frameworks do not discriminate against certain groups (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e). This poses significant challenges as AI system are prone to amplifying biases in relation to gender, ethnicity, minorities, and disadvantaged groups inherent in their training data (Navigli et al., \u003cspan citationid=\"CR90\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), posing a threat, for example, to the validity automated assessment (Aloisi, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Conversely, AI systems can also serve to \u003cem\u003ereduce\u003c/em\u003e educational inequalities by catering to students with special needs (Kasneci et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Yang et al., \u003cspan citationid=\"CR136\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) and learners in deprived regions with little resources (Butgereit \u0026amp; Van Staden, 2023). Thereby, AI may not only be a motivator for changing educational practices, but it could also assist teachers in implementing new models which recognize each student in their individual path towards a flourishing life.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.2.3 Human Flourishing Education and AI: Restoring Meaning to Individual Lives\u003c/h2\u003e \u003cp\u003eMore than prior technologies, AI systems mimic human behavior and decision-making. This poses the fundamental question of distinctly humans characteristics and whether AI systems will increasingly impair human autonomous decision-making and identity (Korseberg \u0026amp; Elken, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e), thereby potentially impacting their ability to lead a flourishing life. This possible loss of purpose and identity in life is contrasted by the potential of AI to render human work more meaningful by allowing humans to focus on more meaningful activities (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e). This fundamental ambivalence is reflected in the educational domain: fears of teacher replacement (Shankar et al., 2025) stand opposed to hopes for augmentation of teacher capabilities (Holstein \u0026amp; Aleven, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2022\u003c/span\u003e); concerns about overreliance on AI systems (Darvishi et al., \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) are contrasted to the opportunity to focus on more complex meaningful activities in class (Siller et al., \u003cspan citationid=\"CR113\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo ensure that AI is used to the benefit of students, teachers and the society at large, a human-centered approach to AI is essential (Bulathwela et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Miao \u0026amp; Holmes, \u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), which \u0026ldquo;ensur[es] human agency and human accountability, and understanding AI\u0026rsquo;s societal impact and implications for citizenship in the era of AI\u0026rdquo; (Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e, p. 23). The development of AI systems should not be centered on the technically possible but on the educationally promising (Barno et al., 2024). Therefore, customized AI tools need to be evaluated before application (Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) to make sure they \u0026ldquo;function effectively while aligning with ethical standards and pedagogical goals\u0026rdquo; (Virvou \u0026amp; Tsihrintzis, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2024\u003c/span\u003e, p. 2).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.3 STEM Education for Human Flourishing\u003c/h2\u003e \u003cp\u003eHuman Flourishing should be seen as an overarching goal rather than a goal of individual lessons or subjects (OECD, \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). However, (formal) education is usually structured in lessons where technology is employed to facilitate the teaching and learning of the respective discipline and reach disciplinary goals (Ergras et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). We therefore narrow our focus to specific disciplines, namely to STEM education.\u003c/p\u003e \u003cp\u003eSTEM Education faces severe challenges which might affect student flourishing in the present and the future. Large-scale international student assessment studies have recently shown that 46% of the 8th grade students do not like learning mathematics and 55% of the 8th grade students report do not feel confident in mathematics (Von Davier et al., \u003cspan citationid=\"CR128\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Similar numbers can be found for physics and chemistry. Moreover, mathematics anxiety has risen in most countries over the past 10 years (OECD, \u003cspan citationid=\"CR95\" class=\"CitationRef\"\u003e2024a\u003c/span\u003e). At the same time, STEM disciplines might be key to ensuring Human Flourishing in future societies. Human Flourishing requires intertwining scientific and moral reasoning, drawing on both adaptive problem solving and ethical decision making (Stevenson, \u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This duality aligns with demands for education in STEM to nurture students\u0026rsquo; reasonable decision-making and critical thinking and encourage active, responsible, democratic citizenship (for mathematics education, see e.g., Geiger et al., \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Siller et al., \u003cspan citationid=\"CR114\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; for science education, see e.g., Yacoubian, \u003cspan citationid=\"CR134\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Thereby STEM education could play a key role in enabling students to contribute to Human Flourishing for their communities and to fair and sustainable futures. In particular, knowledge from diverse STEM disciplines is crucial to understanding (AI) technologies and their societal implications (Siller et al., \u003cspan citationid=\"CR113\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Su (\u003cspan citationid=\"CR119\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) adds to these potentials specifically for mathematics education by arguing that mathematics cultivates virtues connected to Human Flourishing such as play, truth, beauty and justice that are fundamentally rooted in human identity. This situation \u0026ndash; challenges to Human Flourishing in STEM education on the one hand and the potential of STEM education for Human Flourishing on the other \u0026ndash; now coincides with increasingly AI-rich teaching-learning-scenarios in STEM classrooms, which bring about their own tensions, ambivalences and transformative effects (see sections \u003cspan refid=\"Sec5\" class=\"InternalRef\"\u003e2.2.1\u003c/span\u003e to \u003cspan refid=\"Sec7\" class=\"InternalRef\"\u003e2.2.3\u003c/span\u003e). Some of them have been explored in literature reviews on AI in STEM education.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Reviews on AI in STEM Education\u003c/h2\u003e \u003cp\u003eReviews on (generative) AI in STEM education differ in various respects, addressing tensions, ambivalences and transformative effects of AI in STEM education from different angles. For example, Roberts and Mohamed (\u003cspan citationid=\"CR106\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) and Humble (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) focus conduct a SWOT analysis, which identifies \u003cspan type=\"BoldUnderline\" class=\"BoldUnderline\" name=\"Emphasis\"\u003es\u003c/span\u003etrengths, \u003cspan type=\"BoldUnderline\" class=\"BoldUnderline\" name=\"Emphasis\"\u003ew\u003c/span\u003eeaknesses, \u003cspan type=\"BoldUnderline\" class=\"BoldUnderline\" name=\"Emphasis\"\u003eo\u003c/span\u003epportunities and \u003cspan type=\"BoldUnderline\" class=\"BoldUnderline\" name=\"Emphasis\"\u003et\u003c/span\u003ehreats of generative AI in computer science education. Strengths and weaknesses pertain to the characteristics of the AI system, i.e. tasks the AI system is able to complete well or poorly respectively, while opportunities and threats refer to potential uses and outcomes of using AI (Roberts \u0026amp; Mohamed, \u003cspan citationid=\"CR106\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). SWOT analyses originally stem from management (Helms \u0026amp; Nixon, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2010\u003c/span\u003e) but have been used in various other disciplines, including the field of AI in education (e.g., Denecke et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Farrokhnia et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Harris \u0026amp; Kittur (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) take a different approach and identify research themes related to AI in education, suggesting \u0026ldquo;AI Integration\u0026rdquo;, \u0026ldquo;Ethical and Academic Integrity Considerations\u0026rdquo;, \u0026ldquo;AI's Role in Personalized Learning and Assessment\u0026rdquo;, \u0026ldquo;Challenges and Limitations of AI in Education\u0026rdquo; and \u0026ldquo;Future of AI in [\u0026hellip;] Education\u0026rdquo; as key topics addressed in the research. Identifying \u003cem\u003eresearch themes\u003c/em\u003e allows researchers to \u0026ldquo;gain a better understanding of the current landscape and make reasonable predictions about the field\u0026rsquo;s future trajectory\u0026rdquo; (X. Chen et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2024\u003c/span\u003e, p. 17496). It has also been used outside of STEM education (e.g., X. Chen et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Yusuf et al., \u003cspan citationid=\"CR140\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Both approaches highlight issues related to Human Flourishing such as overreliance, ethical concerns or individualization.\u003c/p\u003e \u003cp\u003eThe reviews on AI in STEM education also differ with respect to the disciplines, AI tools and educational levels of interest. While some reviews address mathematics education (Listyaningrum et al., \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Opesemowo \u0026amp; Adewuyi, \u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Pepin et al., \u003cspan citationid=\"CR101\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), engineering education (Chu \u0026amp; Lim, 2023; Filippi \u0026amp; Motyl, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Harris \u0026amp; Kittur, \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) or computer science education (Amos et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Raihan et al., \u003cspan citationid=\"CR103\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Roberts \u0026amp; Mohamed, \u003cspan citationid=\"CR106\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) in particular, others focus on STEM education in general (Loke et al., \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Moreover, Opesemowo and Adewuyi (\u003cspan citationid=\"CR97\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) focus on AI in general, Roberts and Mohamad (2024) address generative AI in particular and Pepin et al. (\u003cspan citationid=\"CR101\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) take an even narrower approach in targeting only ChatGPT. This limitation to generative AI or even ChatGPT is understandable, given its recent surge in public attention and the need to address the potentials and challenges unique to generative AI. Furthermore, the reviews on AI in STEM education differ in the educational levels which are considered; for instance, Harris \u0026amp; Kittur (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) focus on undergraduate engineering education, Chu \u0026amp; Lim (2023) on engineering teacher education and Loke et al. (\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) review STEM pre-service teachers\u0026rsquo; perceptions of AI. If the educational level is not specified but analyzed in the review, articles have found a large share of studies from higher education (Filippi \u0026amp; Motyl, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Raihan et al., \u003cspan citationid=\"CR103\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). This lends to the conclusion that there might be a greater focus on higher education so far, both in the reviews and the original articles on the integration of (generative) AI in STEM education.\u003c/p\u003e \u003cp\u003eIn this review, we therefore focus particularly on \u003cem\u003esecondary\u003c/em\u003e education and the integration of generative AI. Considering that the integration of generative AI in education is in its infancy and research articles are only starting to be published, we broaden the range of potentially eligible research by including all STEM disciplines in the review. Additionally, different from previous reviews, we derive implications for future research regarding Human Flourishing.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Research Questions","content":"\u003cp\u003eThe objective of this review is to present a comprehensive overview of the current research on generative AI in secondary STEM education and to investigate to what extent the current research can serve as a foundation for exploring Human Flourishing and generative AI in STEM education. Specifically, informed by prior reviews of generative AI in STEM education, we will address three research questions:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eWhat are the characteristics of the current research on generative AI in secondary STEM education regarding discipline, publication time, country, journal or conference type, research method, data source, participants, and tools?\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eHow prevalent are different research themes in the current research on generative AI in secondary STEM education?\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003e \u003cem\u003eWhat are the strengths, weaknesses, opportunities and threats (SWOTs) of generative AI for teaching and learning in secondary STEM education?\u003c/em\u003e \u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e"},{"header":"4. Methodology","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Search Strategies and Data Collection\u003c/h2\u003e \u003cp\u003eTo address the research questions, we conduct a scoping review following the PRISMA Extension for Scoping Reviews (Tricco et al., \u003cspan citationid=\"CR123\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). A scoping review aims to \u0026bdquo;determine the scope or coverage of a body of literature on a given topic and give clear indication of the volume of literature [\u0026hellip;] as well as an overview [\u0026hellip;] of its focus\u0026rdquo; (Munn et al., \u003cspan citationid=\"CR89\" class=\"CitationRef\"\u003e2018\u003c/span\u003e, p. 2). It therefore seems particularly suited to identify general characteristics, research themes as well as SWOTs. The process of identification, screening and inclusion is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec13\" class=\"Section3\"\u003e \u003ch2\u003e4.1.1 Identification\u003c/h2\u003e \u003cp\u003eThe literature search was conducted by the first author using five databases \u0026ndash; Web of Science (WoS) Core Collection (1), ERIC (2), Teacher Reference Center (TRC) (3), ArXiv (4) and SCOPUS (5) \u0026ndash; on October 24, 2024 ((1)\u0026ndash;(4)) and November 05, 2024 (5). In addition to the commonly used databases for educational research \u0026ndash; WoS, ERIC, TRC and SCOPUS \u0026ndash; we decided to include ArXiv as a database for two reasons: First, it is a key database for technical research, and the review concerns technical topics. Second, ArXiv publishes preprints, allowing for the inclusion of the latest research results, which is crucial in the rapidly developing field of AI.\u003c/p\u003e \u003cp\u003eIn order to obtain relevant research on generative AI in secondary STEM education, the search string required each record to contain at least one AI-related term, one discipline-related term and one education-related term in its title, abstract or keywords. Alternatively, a record could contain at least one AI-related term and the term \u0026ldquo;technology education\u0026rdquo; or similar combinations\u003csup\u003e2\u003c/sup\u003e. The initial search string was peer-reviewed by an expert on AI in mathematics education according to the PRESS guidelines (McGowan et al., \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). The review resulted in the following final search string for all databases:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e \u003cem\u003e(ChatGPT OR \u0026ldquo;Chat-GPT\u0026rdquo; OR LLM* OR \u0026ldquo;large language model*\u0026rdquo; OR \u0026ldquo;gen* AI\u0026rdquo; OR GenAI OR \u0026ldquo;Generative Artificial Intelligence\u0026rdquo;) AND (((math* OR STEM OR physics OR \u0026ldquo;computer science\u0026rdquo; OR science OR chemistry OR biology OR engineering) AND (education OR school OR college OR universit* OR \u0026ldquo;professional development\u0026rdquo; OR teach* OR student* OR learn* OR curricul*)) OR \u0026ldquo;technology education\u0026rdquo; OR \u0026ldquo;technology teach*\u0026rdquo; OR \u0026ldquo;technology student*\u0026rdquo; OR \u0026ldquo;technology curricul*\u0026rdquo;)\u003c/em\u003e \u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe specifics for individual databases shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e limit the results to English, peer-reviewed records (if allowed as a filter by the respective database) that contain the search string in their title or abstract (or keywords). We do not specify the date of publication; even though generative AI has become more present in the public debate since the release of ChatGPT in 2022, it had existed before, and earlier literature could also inform current and future research.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSpecific search settings in databases\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDatabase\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSpecific search options\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWoS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSearch in \u003cem\u003eTopic\u003c/em\u003e and Language \u003cem\u003eEnglish\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eERIC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSearch in Abstract (\u003cem\u003eAB\u003c/em\u003e) and Title (\u003cem\u003eTI\u003c/em\u003e)\u003c/p\u003e \u003cp\u003eLimiters \u0026ndash; peer reviewed; Language: English; Expanders\u0026nbsp;\u0026ndash; Apply equivalent subjects; Search modes\u0026nbsp;\u0026ndash; Find all my search terms\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTRC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSearch in Abstract (\u003cem\u003eAB\u003c/em\u003e) and Title (\u003cem\u003eTI\u003c/em\u003e)\u003c/p\u003e \u003cp\u003eFilter\u0026nbsp;\u0026ndash; Peer-Reviewed\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSCOPUS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSearch in \u003cem\u003eTITLE-ABS-KEY\u003c/em\u003e and Language \u003cem\u003eEnglish\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eDue to limitations in the Advanced Search mode in ArXiv, the ArXiv API was accessed via a Python (3.8) script to search for all combinations of AI-related, discipline-related and education-related terms. The Python script was used to obtain the ArXiv-IDs of each record, remove duplicates within ArXiv (i.e. not between different databases), retrieve relevant information on the records from ArXiv (title, authors, year, abstract, journal, URL) and save them to a bixtex-file. Citavi7 and Excel-Microsoft365 were used for managing and analyzing the records. The identification process led to a total of 6180 records across all five databases (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) which resulted in 4731 records after removing duplicates.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e4.1.2 Screening and Inclusion\u003c/h2\u003e \u003cp\u003eThe 4731 records identified after removing duplicates were screened for eligibility based on title, abstract and keywords according to the inclusion criteria in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Apart from journal articles, we include conference publications to capture results that are not (yet) published in journals. Preprints allow us to capture the latest research results, which is highly relevant in the rapidly changing field of AI. We further screen the titles, abstracts and keywords for alignment with the three key components of the research questions, i.e. generative AI, secondary education and STEM disciplines.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eInclusion criteria\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"1\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInclusion criteria\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWritten in English\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eJournal article, conference contribution or preprint\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFocusses on generative AI\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFocusses on secondary education\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFocusses on one or more STEM disciplines or STEM as a whole\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe screening process resulted in 243 eligible records for full-text screening. Based on the full-text reports, 183 records were identified as eligible for qualitative analysis according to the inclusion criteria in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Data Analysis\u003c/h2\u003e \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e \u003ch2\u003e4.2.1 (RQ1) General Characteristics\u003c/h2\u003e \u003cp\u003eFor answering the first research question, we categorized the studies according to various dimensions based on deductive category systems for each dimension:\u003c/p\u003e \u003cp\u003e \u003cb\u003eDiscipline.\u003c/b\u003e The discipline addressed in the publications is categorized according to the different STEM disciplines (Science, Technology, Engineering, Mathematics), where Science is further divided into Physics, Chemistry, Biology and Earth Science if indicated in the publication. Moreover, we consider a category \u0026ldquo;STEM\u0026rdquo; if no further subdivision is indicated in the publication (e.g. interviews with STEM teachers).\u003c/p\u003e \u003cp\u003e \u003cb\u003eTime.\u003c/b\u003e As the publications are expected to be concentrated in 2023 and 2024, not only the year, but also the month of publication was identified to allow for a more nuanced analysis. Journal articles were categorized according to the month and year of their publication, conference publication according to the month and year of the conference and preprints according to the month and year of the latest upload on the respective platform (e.g., ArXiv).\u003c/p\u003e \u003cp\u003e \u003cb\u003eCountry.\u003c/b\u003e The publications were categorized according to the countries of the affiliations of all authors.\u003c/p\u003e \u003cp\u003e \u003cb\u003eJournal or conference type.\u003c/b\u003e For journal and conference publications, seven types of journal or conference types were considered based on the self-description of the journal or conference: (1) focusing on education in one STEM discipline, (2) focusing on education in several STEM disciplines, (3) focusing on education in general, (4) focusing on general education and technology, (5) focusing on technology, and (6) other. Note that journals addressing one or several STEM disciplines \u003cem\u003eand\u003c/em\u003e technology, are categorized as (1) or (2) and not as (4).\u003c/p\u003e \u003cp\u003e \u003cstrong\u003ePublication Type and Research method\u003c/strong\u003e \u003cp\u003eWe identify the research method for original research articles as qualitative, quantitative or mixed or multi methods. We further categorize the other publications as theoretical analysis, review, editorial or position paper. Moreover, we categorize publications that do not describe the research method as reports.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eData Source and participants\u003c/strong\u003e \u003cp\u003eWe distinguish the data sources of original research articles as interview, survey, questionnaire, AI-generated content, test, observation, audio/video recording, student work, log data (excluding chatlogs, which are categorized as 4\u0026thinsp;+\u0026thinsp;8) and measurements. In the case of human participants, we distinguish between students, teachers, pre-service teachers (PSTs) and others.\u003c/p\u003e \u003c/p\u003e \u003cp\u003e \u003cstrong\u003eTools\u003c/strong\u003e \u003cp\u003eWe differentiate between standard generative AI systems (e.g., ChatGPT or Gemini) and non-standard AI systems, e.g. self-developed, fine-tuned tools or retrieval augmented generation\u003csup\u003e3\u003c/sup\u003e (RAG) enhanced systems.\u003c/p\u003e \u003c/p\u003e \u003cp\u003eThe complete coding scheme is provided in the supplementary material.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003e4.2.2 (RQ2) Research Themes\u003c/h2\u003e \u003cp\u003eWe conducted a structuring qualitative content analysis of the full-text publications (Kuckartz \u0026amp; R\u0026auml;diker, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) to address the second research question. Our deductive approach included the development of main categories, which we used to inductively develop subcategories for on the most frequent main categories (Kuckartz, \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The main categories were developed by collecting various articles that investigate research themes for AI in education (see section \u003cspan refid=\"Sec9\" class=\"InternalRef\"\u003e2.4\u003c/span\u003e) and synthesizing the different category systems from these contributions. The synthesized category system included nine categories:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTheoretical frameworks\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eEvolution, future directions, trends \u0026amp; philosophical considerations\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePros \u0026amp; cons\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTool development\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003ePerformance\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eEthical considerations \u0026amp; academic integrity\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eAwareness, attitude \u0026amp; acceptance\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eApplication \u0026amp; impact\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eEducational research\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eThe category systems including category descriptions can be found in the supplementary material. The full texts were considered as a coding unit. In principle, we allowed double coding, but only if the paper was divided into separate parts with different research goals; otherwise, we assigned only one research theme according to the dominant theme of the publication.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section3\"\u003e \u003ch2\u003e4.2.3 (RQ3) SWOT Analysis\u003c/h2\u003e \u003cp\u003eFor addressing the third research question, we conducted a SWOT analysis of the records using structuring qualitative content analysis (Kuckartz \u0026amp; R\u0026auml;diker, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The four components of SWOT \u0026ndash; strengths, weaknesses, opportunities and threats \u0026ndash; were used as main categories developed by deductive category formation. A precise description of these categories \u0026ndash; similar to prior research (see section \u003cspan refid=\"Sec9\" class=\"InternalRef\"\u003e2.4\u003c/span\u003e) \u0026ndash; is shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Subcategories were developed inductively to ensure that the categories can be specific to the STEM disciplines.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eMain categories of SWOT analysis\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCategory\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStrength\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eProperty of generative AI systems that is conducive to the teaching and/or learning of the respective STEM discipline(s) in secondary education\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWeakness\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eProperty of generative AI systems that is obstructive to the teaching and/or learning of the respective STEM discipline(s) in secondary education\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOpportunity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eInfluence of generative AI systems on teachers and students that fosters the teaching and/or learning of the respective STEM discipline(s) in secondary education\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eThreat\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eInfluence of generative AI systems on teachers and students that inhibits the teaching and/or learning of the respective STEM discipline(s) in secondary education\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eIn order to obtain evidence-based results, we included only empirical research in the SWOT analysis. Sentences or parts of sentences in the abstract and conclusion of each publication were used as coding units. Abstract and conclusion are meant to summarize the findings and identify their main implications and are therefore particularly suitable for the identification of SWOTs. Instead, coding sentence-wise in the \u003cem\u003ewhole\u003c/em\u003e publication would pose the risk of losing sight of the context and overrepresenting individual statements instead of the main message of the paper. The analysis was conducted using MaxQDA24.\u003c/p\u003e \u003cp\u003eThe paraphrases of all passages in a main category were collected. Using paraphrases for inductive category development is inspired by Mayring\u0026rsquo;s (\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e2014\u003c/span\u003e) summarizing content analysis and applied here for two reasons: first, the large number of passages makes verbatim collection difficult, and second, paraphrasing incorporates contextual information (e.g. about the discipline referred to). Inductive development of subcategories was applied for each of the four main categories based on the paraphrases.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"5. Results","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Characteristics of the current research on AI in secondary STEM education\u003c/h2\u003e \u003cp\u003eRegarding the distribution of publications across disciplines, mathematics is the most frequently covered discipline (39%), followed by physics (24%) and science (19%). All other disciplines as well as STEM as a whole are addressed in 10% or less of the publications (see Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Moreover, biology, chemistry and earth science are addressed in conjunction with other disciplines in the majority of publications where they occur.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRegarding the month and date of publication, an overall increase can be noted (see Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). There is a peak in July 2024, which can be mainly attributed to several conferences on AI in education at this time. The downwards trend after July 2024 should not be overinterpreted, as a drop from July to August also occurred in 2023 and for Oktober 2024, not all results are included due to a retrieval date for this review. Furthermore, it can be noted that one publication was published in May 2021, while all other papers are published after 2023, i.e. after the release of ChatGPT. Note that four publications were not included in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, as they used to be preprints at the time of identification of records but had been published at the time of full-text retrieval.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe geographical distribution reveals that the United States are by far the most active country with the most contributions, accounting for 30% of the global publications, followed by China (8%), Germany (8%) and the United Kingdom (5%) and Brazil (5%). Overall, 73% of the affiliations are situated in the Global North and only 27% in the Global South. The distribution across different disciplines shows that the concentration in the Global North is particularly strong for technology education and science education and less than average for all other disciplines. The concentration in the Global North is even more pronounced for conference publications, where 88% are contributions from the Global North and 48% from the United States. Conversely, for journal articles, the geographical distribution is a bit more levelled out, with 15% from the United States, followed by Germany (13%), China (9%), Brazil (6%), Australia (6%) and South Africa (4%).\u003c/p\u003e \u003cp\u003eFor the journal or conference type, it should first be noted that 14% of the papers are preprints and thus not considered here. Of the remaining contributions (33% conference publications, 53% journal articles), 17% are published in journals or at conferences targeting one STEM discipline, 14% in journals or on conferences focusing on several STEM disciplines. Another 10% of journals and conferences are dedicated to education in general. More than half of journals and conferences either focus on technology (18%) or on education in conjunction with technology (34%). The remaining 7% do not belong to any of these categories (see Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). For individual disciplines, papers from the sciences are more frequently published in STEM-specific journals and conferences, while this is rare for mathematics. Mathematics papers are predominantly published in journals and conferences on education and technology.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRegarding the methodology, we consider research method, data source, and participants. Out of the 183 publications, 76% present original research articles, which include full data collection, analysis and presentation of results. They are almost equally split across qualitative (33%), quantitative (30%) and mixed/multi methods (37%) research designs. 16% of the publications are reports without a description of the methodology, describing for example a teaching unit or the development process of an AI tool. The remaining publications are reviews (2%), editorials (1%), position papers (1%) and theoretical analyses (4%) (see Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Concerning individual disciplines, the share of original research is particularly high for biology, earth science, technology and mathematics. Moreover, mathematics, physics, and science reflect the almost equal split across qualitative, quantitative and mixed-methods, while chemistry is dominated by qualitative research and STEM is dominated by quantitative and mixed-methods designs. The number of reports is particularly high in chemistry and rather low in mathematics.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRegarding the data source (see Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e), half of the original research publications collect, analyze and present AI-generated content. For papers investigating human participants, using surveys, interviews and students work is most common, often in conjunction with each other or with AI-generated content. Moreover, the most commonly assessed group of human participants are students, followed by teachers, pre-service teachers and others (e.g. parents, teacher educators).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWith regards to the tools, 9% of the publications use (sometimes among others) fine-tuned or self-developed AI systems or integrate retrieval augmented generation (RAG). The remaining 91% resort to existing AI systems, predominantly to ChatGPT.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Research Themes\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e: \u003cem\u003eResearch themes overall (black) and divided by discipline (colored); bars are higher for disciplines because publications addressing multiple disciplines are counted multiple times for the respective research theme\u003c/em\u003e\u003c/p\u003e \u003cp\u003eThe results show that \u003cem\u003eperformance\u003c/em\u003e is the most frequent research theme (35%), followed by \u003cem\u003etool development\u003c/em\u003e (22%) and \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e (19%) of AI tools. All other themes are rather rare. The different themes are approximately equally distributed across the disciplines. In particular, \u003cem\u003eperformance\u003c/em\u003e is the most frequent category for all disciplines except for the three infrequent ones (\u003cem\u003etechnology\u003c/em\u003e, \u003cem\u003eengineering\u003c/em\u003e and \u003cem\u003eSTEM\u003c/em\u003e). Some disciplines are not represented across all research themes: Notably, the most frequent discipline \u0026ndash; mathematics \u0026ndash; does not appear for \u003cem\u003eethical considerations \u0026amp; academic integrity\u003c/em\u003e. Moreover, mathematics is represented in the \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e theme almost as frequently as physics, although the overall number of mathematics-related publications is much higher. Interestingly, for technology education, \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e is the most common theme.\u003c/p\u003e \u003cp\u003eWe take a closer look at the research themes by (1) presenting subcategorizations of the three largest research themes and (2) considering the development of research themes over time, in relation to data sources and participants as well as the tools used across the research themes.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e shows that the kind of \u003cem\u003eperformance\u003c/em\u003e assessed ranges from solving mathematical tasks \u0026ndash; in competition problems (e.g., Teegavarapu \u0026amp; Sanghvi, 2023), official exams (e.g., Dao \u0026amp; Le, 2023), and other questions (e.g., Schorcht, Buchholtz, et al., 2024) \u0026ndash; to engaging in pedagogical considerations by planning lessons (e.g., Hu et al., 2024), tutoring and dialogs (e.g., Gregorcic \u0026amp; Pendrill, 2023), or generating feedback and assessment (e.g., Bewersdorff et al., 2023). Similar categories can be identified for \u003cem\u003etool development\u003c/em\u003e, which shows that apart from \u003cem\u003eassessing\u003c/em\u003e the performance of generative AI in pedagogical activities, there are efforts to \u003cem\u003eimprove\u003c/em\u003e the performance of generative AI in these areas. Moreover, almost a quarter of the publications in the \u003cem\u003etool development\u003c/em\u003e category is concerned with the compilation of datasets, for example for multimodal mathematical reasoning (e.g., Z.-Z. Li et al., 2024) or high school physics questions (e.g., Anand et al., 2023).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFor \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e, it can be found that 33% of the publications focus on the impact of using generative AI on cognitive variables (e.g., Dasari et al., 2024; Ghazali et al., 2024; T. Li et al., 2024) such as performance in STEM disciplines or cognitive load. Another 8% address effects on non-cognitive variables (e.g., intrinsic motivation or self-confidence) and 17% focus on both (e.g., Canonigo, 2024; Lu et al., 2024). The remaining publications describe the behavior of participants while engaging in activities with AI (e.g., the behavior of students working with GeoGebra using ChatGPT in Yunianto et al., 2024) (17%), describe an activity about dealing with AI and assess student performance in it (e.g. with AI-generated science texts in Cheung et al., 2024) (8%) or solely describe teaching or learning activities in which generative AI (Pavlova, 2024) (17%).\u003c/p\u003e \u003cp\u003eWe further consider the development of the research themes over time by three-month units to avoid rapid fluctuations. Figure\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e (left) shows the prevalence of research on the \u003cem\u003eperformance\u003c/em\u003e of AI tools at all times, as well as a rapid increase of research on tool development since the end of 2023. In the beginning of 2024, the \u003cem\u003etool development\u003c/em\u003e category surpasses the \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e category. For the remaining categories, the total number of publications is too low to draw reasonable conclusions across time.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRelating the participants to research themes reveals some noticeable patterns (see Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e, right). Unsurprisingly, most publications on \u003cem\u003etool development\u003c/em\u003e and \u003cem\u003eperformance\u003c/em\u003e of AI tools do not involve humans. However, \u003cem\u003eif\u003c/em\u003e tools are tested with humans, participants are mostly students. Moreover, for the \u003cem\u003eawareness, attitude \u0026amp; acceptance\u003c/em\u003e category, participants are mostly teachers. Conversely, when impact and application of AI tools is assessed, this is very rarely done with teachers \u0026ndash; the most frequent participant groups are students, followed by pre-service teachers.\u003c/p\u003e \u003cp\u003eRegarding the use of standard AI tools versus non-standard AI tools (i.e. self-developed tools, fine-tuned tools or tools using RAG), it can be noted that non-standard tools are used only in 9% of the publications, mostly for the research themes \u003cem\u003etool development\u003c/em\u003e and \u003cem\u003eperformance\u003c/em\u003e. Publications in other research themes (almost) exclusively use standard tools.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e5.3 SWOT Analysis\u003c/h2\u003e \u003cp\u003eThe SWOT analysis summarizes the strengths, weaknesses, opportunities and threats of generative AI in secondary STEM education. The results are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e. As elaborating on each factor for all STEM disciplines would go beyond the scope of this review, we focus on mathematics as the most prominent discipline in the review and at times include examples from other disciplines if only few or no mathematical examples are available.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003e5.3.1 Strengths\u003c/h2\u003e \u003cp\u003eGenerative AI systems are able to accurately solve STEM tasks in official/national exams such as the NAEP mathematics questions (Wei, 2024). They excel particularly at basic (Dao \u0026amp; Le, 2023), knowledge-level mathematics questions (Guler et al., 2024) and basic mathematical modelling tasks (Spreitzer et al., 2024). Strengths can be found especially in the areas of algebra (Teegavarapu \u0026amp; Sanghvi, 2023), numbers (Dao \u0026amp; Le, 2023; Teegavarapu \u0026amp; Sanghvi, 2023; Wei, 2024) and probability and statistics (Vank\u0026uacute;š, 2024). Beyond solving tasks, generative AI systems can provide explanations of solutions (Daher \u0026amp; Gierdien, 2024; Ergene \u0026amp; Ergene, 2025) and offer explanations for their scoring of student answers (Lee et al., 2024). Moreover, generative AI systems stand out for their ability to provide mathematics tutoring in various languages (Butgereit \u0026amp; Van Staden, 2023). Regarding educational materials, generative AI systems can create contextually relevant questions (van Pham et al., 2024), adapt mathematics questions to diverse ability levels (Rouzegar \u0026amp; Makrehchi, 2024) and effectively tailor the language of mathematics tasks to different learners (Norberg et al., 2024). They further stand out for their ability to create clear and organized lesson plans with well-defined instructional objectives, organization, methods and strategies, particularly in the areas of functions and statistics (Hu et al., 2024). Generative AI systems can analyze discourse in mathematics classrooms similar to human coders (Long et al., 2024) and imitate mathematics students (Drushlyak et al., 2024; Lu et al., 2024; Rouzegar \u0026amp; Makrehchi, 2024). An overarching strength of generative AI systems is their customizability for educational purposes: This can improve the generation of questions (R. Li et al., 2024), distractors and error labels (Fernandez et al., 2024) in mathematics multiple-choice questions, enhance the performance in mathematics question answering (Levonian et al., 2023; Zhang et al., 2024) and lead to more effective simulations of students (Jin et al., 2024; Sonkar et al., \u003cspan citationid=\"CR116\" class=\"CitationRef\"\u003e2024\u003c/span\u003e) \u0026ndash; for example in collaborative mathematical modelling (Yue et al., 2024) \u0026ndash; as well as to more accurate scoring and evaluation (Latif \u0026amp; Zhai, 2024; Nicula et al., 2023). Other strengths of generative AI for STEM education pertain to the avoidance of bias in some cases (Cooper \u0026amp; Tang, 2024; Kunz \u0026amp; Steffen, 2024) and to their large knowledge base, providing interdisciplinary knowledge (dos Santos, 2023).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section3\"\u003e \u003ch2\u003e5.3.2 Weaknesses\u003c/h2\u003e \u003cp\u003eA major weakness of generative AI systems is their tendency to produce wrong information and make mistakes (Butgereit \u0026amp; Van Staden, 2023; Daher \u0026amp; Gierdien, 2024; Ergene \u0026amp; Ergene, 2025; Guler et al., 2024; Prihar et al., 2023; Ribeiro et al., 2024; Spreitzer et al., 2024; Taani \u0026amp; Alabidi, 2024). Difficulties arise particularly for complex problems (Dao \u0026amp; Le, 2023; Teegavarapu \u0026amp; Sanghvi, 2023; Wei, 2024) and problems that require nuanced understanding (Spreitzer et al., 2024), formal reasoning (Parra et al., 2024) or induction (Dasari et al., 2024). Generative AI systems struggle with spatial reasoning and geometry (Dao \u0026amp; Le, 2023; Guler et al., 2024; Wardat et al., 2023; Wei, 2024), calculus (Dao \u0026amp; Le, 2023) as well as combinatorics and probability (Guler et al., 2024; Teegavarapu \u0026amp; Sanghvi, 2023). Problems in geometrical tasks align with a general difficulty of generative AI systems with graphical input and output \u0026ndash; they struggle for example in visual proofs (Schorcht, Baumanns, et al., 2024) and multimodal mathematics reasoning (Z.-Z. Li et al., 2024). Also, performance may be worse for non-English mathematics tasks (Dao \u0026amp; Le, 2023; Parra et al., 2024). Apart from \u003cem\u003ecommitting\u003c/em\u003e mathematical errors, generative AI systems also struggle with anticipating (Feng et al., 2024), understanding (McNichols et al., 2024) and categorizing (Yan et al., 2024) (common) mathematical errors. They further show difficulties in identifying student reasoning for flawed solutions (Liu et al., 2024) and differentiating mathematical errors from student insecurities (Kakarla et al., 2024). As generative AI systems are based on huge amounts of text but not on real-world experiences, they may lack real-world knowledge when creating mathematics lesson plans, for example on the curriculum, mathematical culture and suitable technology integration (Baytak, 2024; Egara \u0026amp; Mosimege, 2024; Hu et al., 2024). Sometimes, the AI-generated educational materials \u0026ndash; distractors (Feng et al., 2024; McNichols et al., 2023), guidance prompts (Dasari et al., 2024), generated questions (Vank\u0026uacute;š, 2024) and hints (Gattupalli et al., 2023) \u0026ndash; do not keep up with human ones or are less preferred by experts. Other weaknesses of generative AI for STEM education concern the issue of bias, for example concerning STEM job recommendations regarding gender (Due et al., 2024) or the negative perception of mathematics and STEM (Abramski et al., 2023).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003e5.3.3 Opportunities\u003c/h2\u003e \u003cp\u003eWe differentiate between opportunities for students, for teachers and for pre-service teachers.\u003c/p\u003e \u003cp\u003eGenerative AI systems can assist students in learning and understanding mathematics (Canonigo, 2024; Daher \u0026amp; Gierdien, 2024; Egara \u0026amp; Mosimege, 2024; Schorcht, Buchholtz, et al., 2024; Yunianto et al., 2024). For example, they can translate mathematical language to natural language (Ribeiro et al., 2024) and integrate various modalities (Schorcht, Baumanns, et al., 2024), leading to an increase in (conceptual) mathematical understanding (Canonigo, 2024; Egara \u0026amp; Mosimege, 2024). Similarly, generative AI systems can help students solve (basic) mathematics tasks and thereby lead to increased student performance (Henkel, Horne-Robinson, Kozhakhmetova, et al., 2024; Norberg et al., 2024; Rouzegar \u0026amp; Makrehchi, 2024). Moreover, generative AI systems can promote self-directed, (inter)active learning, for example in mathematics flipped classroom settings (Pavlova, 2024). The use of generative AI can also contribute to the students\u0026rsquo; development of critical thinking and reflection skills, for example when students categorize errors of AI systems in geometry (Parra et al., 2024) or analyze mathematical mistakes made by these systems (Bellettini et al., 2023). Generative AI can further promote problem-solving activities and computational thinking when students debug AI generated GeoGebra commands (Yunianto et al., 2024) or create mathematical visualizations using generative AI (Schorcht, Baumanns, et al., 2024). If students solve mathematical tasks collaboratively with an AI system (Daher \u0026amp; Gierdien, 2024) or engage in AI-guided reward-based learning paths (Singh et al., 2024), these systems have the potential to enhance engagement and collaboration in mathematics. Furthermore, they promote inclusivity, democracy and equality, for example by assisting teachers in creating specialized math practice for disabled students (Lin \u0026amp; Riccomini, 2024) and making mathematics education accessible in lower and lower-middle income countries (Henkel, Horne-Robinson, Kozhakhmetova, et al., 2024). All in all, the use of generative AI systems in mathematics education can positively influence affective variables by increasing self-efficacy, confidence (Canonigo, 2024) and self-assurance (Yunianto et al., 2024) in mathematics.\u003c/p\u003e \u003cp\u003eGenerative AI can assist teachers with lesson planning (Baytak, 2024; Hashem et al., 2023; Taani \u0026amp; Alabidi, 2024) and serve as reference for teachers creating mathematics explanations (Prihar et al., 2023) and tips (Jia et al., 2024). Generative AI can thereby reduce teachers\u0026rsquo; workload and increase their teaching efficiency (Egara \u0026amp; Mosimege, 2024; Hashem et al., 2023; Hu et al., 2024). Apart from the creation of educational resources, generative AI systems can help teachers with assessment and feedback. They may analyze students\u0026rsquo; responses and weaknesses in mathematics and science (Taani \u0026amp; Alabidi, 2024) and encourage open-ended, process-oriented assessment formats (Henkel, Horne-Robinson, Dyshel, et al., 2024). Moreover, generative AI systems allow teachers to analyze and practice their own teaching by generating feedback for their teaching (Barno et al., 2024), particularly regarding socio-emotional learning (Han et al., 2024).\u003c/p\u003e \u003cp\u003ePre-service teachers are on the one hand themselves learners of mathematical content in university courses where generative AI can assist them in completing tasks and essays (Dasari et al., 2024; Vank\u0026uacute;š, 2024), leading to higher performance if used in conjunction with a human teacher (Dasari et al., 2024). On the other hand, pre-service teachers need to develop teaching skills. Using generative AI systems, they can practice finding and arguing mathematical errors (Drushlyak et al., 2024), which may lead to higher teacher self-efficacy (Lu et al., 2024), increased critical thinking (Drushlyak et al., 2024) and increased higher order thinking (Lu et al., 2024).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003e5.3.4 Threats\u003c/h2\u003e \u003cp\u003eWe differentiate between threats for students, for teachers and for pre-service teachers.\u003c/p\u003e \u003cp\u003eUsing generative AI in mathematics classrooms can hinder students\u0026rsquo; learning and understanding of mathematics as the system may give immediate answers (Singh et al., 2024) and reinforce or even introduce misconceptions (Daher \u0026amp; Gierdien, 2024; Parra et al., 2024). Generative AI systems may further impede the development of critical thinking and reflection (Dasari et al., 2024; Shankar et al., 2025), independent problem-solving and independent analysis and evaluation of information as students may overly rely on AI systems (Shankar et al., 2025). Other threats of generative AI in STEM education pertain to the reinforcement of stereotypes (Cooper \u0026amp; Tang, 2024; Due et al., 2024), ethics and data security (Latif \u0026amp; Zhai, 2024; Shankar et al., 2025) as well as dishonesty and misbehavior of students (Garofalo \u0026amp; Farenga, 2025; Yeadon \u0026amp; Hardy, 2024).\u003c/p\u003e \u003cp\u003eFor teachers, the potential marginalization or replacement of human teachers\u0026rsquo; involvement in the learning process and of the subtleties of their expertise presents a major threat (Barno et al., 2024; Vasconcelos \u0026amp; Dos Santos, 2023). The adoption of generative AI systems may disadvantage teachers with less digital competencies (Shankar et al., 2025), replace teachers as well as traditional teaching and grading by automation (Latif \u0026amp; Zhai, 2024) and focus too much on the technologically possible and not on the educationally promising in mathematics teacher professionalization (Barno et al., 2024).\u003c/p\u003e \u003cp\u003ePre-service teachers may suffer from decreased performance in mathematics when only learning with generative AI systems (i.e. without a teacher) (Dasari et al., 2024). They may also experience a lower quality of educational materials regarding the clarity of tasks in physics (K\u0026uuml;chemann et al., 2023) and the design of STEM teaching units (Z. Li \u0026amp; Ironsi, 2024).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"6. Discussion","content":"\u003cp\u003eEducation for Human Flourishing strives to enable all students to live up to their potential in life by fostering a sense of purpose, meaning and autonomous decision-making. Recent technological developments call for an exploration of the role of (generative) AI regarding Human Flourishing in educational settings including secondary STEM education. To this end, this article has reviewed the current research on generative AI in secondary STEM education considering general characteristics, research themes and SWOTs. Based on these results, we are going to discuss to what extent the current literature provides a foundations for exploring the interaction of Human Flourishing and generative AI in secondary STEM education, discussing the findings under the three ideas proposed by the OECD (2024) \u0026ndash; \u003cem\u003eBroadening human capabilities\u003c/em\u003e, \u003cem\u003edeveloping new models for the future\u003c/em\u003e, and \u003cem\u003erestoring meaning to individual lives\u003c/em\u003e \u0026ndash; each of them in relation to secondary STEM education. Based on these elaborations, we outline potentials for future research.\u003c/p\u003e \u003cdiv id=\"Sec28\" class=\"Section2\"\u003e \u003ch2\u003e6.1 Broadening Human Capabilities in STEM Education\u003c/h2\u003e \u003cp\u003eIncreasingly advanced AI systems underline the need for a shift from mostly cognitive educational goals to socio- and meta-cognitive objectives in STEM education, focusing on a holistic view of the student (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e). However, the findings of this review show that the effect of using generative AI on cognitive variables is investigated more frequently than on non-cognitive variables. Despite the apparent emphasis on cognitive aspects, the SWOT analysis draws a more holistic picture, revealing opportunities for learners\u0026rsquo; motivation, confidence and self-efficacy, their engagement and collaboration in STEM disciplines as well as the advancement of inclusivity and equality in STEM teaching and learning. These opportunities provide initial evidence that generative AI might be helpful to address challenges to Human Flourishing in STEM education, such as low student confidence and increasing anxiety (OECD, \u003cspan citationid=\"CR95\" class=\"CitationRef\"\u003e2024a\u003c/span\u003e; Von Davier et al., \u003cspan citationid=\"CR128\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). They further highlight that AI could provide an opportunity to foster capabilities that enable students to uphold circumstances that allow flourishing throughout society, such as critical thinking and active engagement as responsible citizens (Geiger et al., \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Yacoubian, \u003cspan citationid=\"CR134\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThere is a need for further research that comprehensively examines the impact of generative AI on students, including metacognitive and socio-emotional factors, to strengthen initial research efforts regarding the associated opportunities.\u003c/p\u003e \u003cp\u003eThe need for broader and more holistic human capabilities places increased emphasis on human ethical decision-making as opposed to automatic choices (Karakuş et al., \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Stevenson, \u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). However, current research shows that there is a limited body of literature addressing ethical considerations and academic integrity, particularly in mathematics education. However, the SWOT analysis shows that various publications mention the threat of misbehavior by students, ethics and data security as well as the reinforcement of stereotypes through generative AI systems. There is a need to explore ways to mitigate these issues in STEM education in order to ensure that the technology promotes inclusivity and democratization (Ergras et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) and fosters ethically sound educational practices (Virvou \u0026amp; Tsihrintzis, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThere is a need for publications that focus on ethical consideration in STEM education in depth.\u003c/p\u003e \u003cp\u003eBy leveraging empirically-based opportunities and mitigating threats, STEM education could play a significant role in fostering Human Flourishing as it provides opportunities to intertwine ethical and scientific reasoning, fostering adaptive problem solving in conjunction with ethical decision-making (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e; Stevenson, \u003cspan citationid=\"CR118\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Given this pivotal role of STEM, our review has included only publications concerning STEM education in some regard. The findings revealed that less than a third of included journal articles and conference contributions are published in STEM education journals or conferences; in turn, almost half of the papers are published in general education journals or conferences, mostly in conjunction with technology. This lends to the conclusion that research on generative AI in STEM education focuses more on pedagogy and technology than on the STEM disciplines themselves. This could complicate drawing targeted conclusions for STEM teaching and learning.\u003c/p\u003e \u003cp\u003eGiven the unique potentials and challenges for Human Flourishing in the STEM disciplines, future research should focus more strongly on discipline-specific applications and implications of generative AI in STEM education in order to derive targeted decisions for STEM classrooms.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec29\" class=\"Section2\"\u003e \u003ch2\u003e6.2 Developing New Models for the Future of STEM Education\u003c/h2\u003e \u003cp\u003eEnhanced capabilities of generative AI systems present challenges to existing \u0026ndash; organizational, political and societal \u0026ndash; models in education, calling for new approaches and solutions, particularly concerning teacher education (Feldman-Maggor et al., 2025; Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Ng et al., \u003cspan citationid=\"CR91\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHowever, this review identified only very few publications that focus on theoretical frameworks for teaching and learning with generative AI in secondary STEM education. While there are diverse overarching frameworks, for example on AI-specific TPACK (Mishra et al., \u003cspan citationid=\"CR87\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), they do not specifically focus on STEM. This reflects the tendency for publications to focus rather generally on education (see section \u003cspan refid=\"Sec28\" class=\"InternalRef\"\u003e6.1\u003c/span\u003e). At the same time, AI provides an opportunity to assist the development of new models (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e) by helping STEM teachers improve their own teaching through advanced discourse analysis tools and AI-generated suggestions for improvements (see SWOT analysis).\u003c/p\u003e \u003cp\u003eFuture research should suggest STEM-specific frameworks on the integration of AI in education and also empirically validate them in order to develop new solutions in an age of generative AI.\u003c/p\u003e \u003cp\u003eWhen new solutions are developed, it is crucial that they do not discriminate against minorities and ensure that also marginalized voices are heard (OECD, \u003cspan citationid=\"CR94\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). This is particularly relevant with regard to Human Flourishing as different cultures may conceptualize a flourishing life differently (Curren et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; de Ruyter et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). However, we found a concentration of publications in the Global North, especially in the United States. This is particularly unfortunate in the field of generative AI, as generative AI systems demonstrate linguistic versatility and a potential to promote inclusivity, democracy and equality in STEM education (see SWOT analysis). Research situated in the Global South does indeed show promising results of AI-assisted mathematics tutoring in under-resourced contexts (Butgereit \u0026amp; Van Staden, 2023). A more diverse authorship could also decrease the threat of bias and reduce stereotypes in generative AI systems, as more diverse data could be used for future training. At least, there seems to be some more balance regarding the geographical distribution for journal articles, with contributions from Brazil, South Africa and China in particular.\u003c/p\u003e \u003cp\u003eThe global research community should foster more equal participation, particularly for conferences, in order to explore how generative AI in STEM education can enhance Human Flourishing everywhere.\u003c/p\u003e \u003cp\u003eThe need for new models, particularly approaches concerning teacher-AI-collaboration in STEM education, raises the question whether a SWOT analysis could be used in this regard. The potential of SWOT analysis consists in providing a comprehensive overview of internal and external factors influencing the successful adoption of generative AI in education. Considering the complexity of the adoption of new technology and the interconnectedness of internal and external factors, it is not sufficient to regard the four components of SWOT in isolation. Moreover, SWOT analyses are not an end in themselves, but a starting point for strategic planning (Helms \u0026amp; Nixon, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2010\u003c/span\u003e). We therefore focus on how teachers can harness the opportunities of generative AI to assist them with the creation of educational materials and the assessment and feedback in STEM education, thereby reducing teacher workload, while mitigating the threat of teacher replacement and the lack of involvement in the learning process (see Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe current strengths of generative AI provide an indication of what teachers could use generative AI systems for, while the weaknesses need to be compensated by the teacher. Teachers could use generative AI for finding suitable contexts for tasks and easily generate examples, while it is up to the teacher to check the correctness, especially for complex, counterintuitive reasoning tasks and tasks involving graphical input and output. Generative AI may be able to provide clear structure for lesson plans or other teaching activities, to which teachers then add their real-world knowledge. Generative AI can be used to further adapt the materials to different students\u0026rsquo; needs and change formulations or even language. Teachers should watch out that no biases are perpetuated by the AI system and intervene if necessary. Regarding assessment and feedback, generative AI, especially when customized for educational purposes, may provide a starting point for scoring, excelling at providing explanations for its decisions. Teachers may then check the scoring, watching out for typical student errors and misconceptions.\u003c/p\u003e \u003cp\u003eIt is crucial to highlight that the presented strategy is not fixed, but subject to dynamic changes as generative AI systems evolve and new research results become available; for instance, current weaknesses that need to be compensated by the teacher, might become strengths in the future (or vice versa), demanding different teacher reactions.\u003c/p\u003e \u003cp\u003eFuture research could implement the suggested strategy in STEM teacher professionalization and evaluate whether it actually reduces teacher workload and increases efficiency, while maintaining the central role of the teachers in the educational process.\u003c/p\u003e \u003cp\u003eWith regards to Human Flourishing, the suggested strategy aims to maintain teacher agency and autonomous decision-making. It selectively delegates tasks to a generative AI system while the teacher remains accountable for final choices on educational materials and student assessment. By balancing human and algorithmic decision-making (Merzifonluoglu \u0026amp; Gunes, \u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), it strives to allow teachers to enhance and not reduce the purposefulness and meaningfulness of their work (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e). This anticipates the next section on \u003cem\u003erestoring meaning to individual lives\u003c/em\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec30\" class=\"Section2\"\u003e \u003ch2\u003e6.3 Restoring Meaning to Individual Lives in STEM Education\u003c/h2\u003e \u003cp\u003eIncreasingly human-like AI systems bear the risk of challenging human identity, agency and autonomous decision-making. It is therefore crucial to adopt a human-centered mindset where the impact of generative AI on humans, the society and the environment is of primary importance. This requires empirical research on the application and impact of generative AI in teaching and learning scenarios in real STEM classrooms (Kong \u0026amp; Yang, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe examination of research methods in the review shows that the vast majority of studies are empirical. However, the primary emphasis of these studies is on the evaluation of AI-generated content investigating the performance of generative AI systems in STEM education or developing and testing customized tools for STEM education. These results lend to the conclusion that the technically possible might be more strongly focused on than the pedagogically promising (Barno et al., 2024). The trend towards tool development instead of the assessment of impact and application is even slightly increasing over time, with \u003cem\u003etool development\u003c/em\u003e surpassing \u003cem\u003eimpact \u0026amp; application\u003c/em\u003e in the beginning of 2024. These results suggest an increasingly technocentric approach, which does not \u0026ldquo;put[\u0026hellip;] the learner at the epicentre of the design of the tools\u0026rdquo; (Bulathwela et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2024\u003c/span\u003e, p. 11). Also, newly developed or customized AI systems are rarely tested with human participants, exposing students and teachers to unexpected risks regarding their cognitive and affective development and well-being (Miao \u0026amp; Cukurova, \u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Vice versa, the application and impact of generative AI in STEM education are never examined using fine-tuned or self-developed tools or tools using RAG. These results suggest a gap between technical development and the actual implementation.\u003c/p\u003e \u003cp\u003eFuture research should place stronger emphasis on the investigation of human application of generative AI in STEM education as well as the impact on students, teachers and pre-service teachers to adopt a human-centered approach to AI in STEM education. This should include increased efforts for cooperation between technical development and practical assessment.\u003c/p\u003e \u003cp\u003eFor the research themes that primarily involve human participants \u0026ndash; \u003cem\u003eawareness, attitude \u0026amp; acceptance\u003c/em\u003e and \u003cem\u003eapplication \u0026amp; Impact\u003c/em\u003e \u0026ndash; teachers constitute the largest group of participants in the theme of \u003cem\u003eawareness, attitude \u0026amp; acceptance\u003c/em\u003e, while research regarding the \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e of generative AI by teachers is largely absent. This tendency counteracts the demand that \u0026ldquo;future research should strive to study these [AI] systems in real-world contexts, with authentic tasks and relevant human experts.\u0026rdquo; (Holstein \u0026amp; Aleven, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2022\u003c/span\u003e, p. 245f). It could also hinder a human-centered approach to AI integration in education (Bulathwela et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Miao \u0026amp; Holmes, \u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The lack of studies for \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e involving teachers results in the lack of nuanced opportunities and threats for teachers in the SWOT analysis. Furthermore, the performance and competence of teachers is rarely assessed, in contrast to the frequent assessment of students, as evidenced by the subcategory of \u003cem\u003eapplication \u0026amp; impact\u003c/em\u003e. However, to engage with generative AI, teacher competencies are of pivotal importance (Mishra et al., \u003cspan citationid=\"CR87\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Ng et al., \u003cspan citationid=\"CR91\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThere is a need for research that goes beyond teachers' attitudes and opinions to investigate actual integration practices and competencies. This could facilitate the identification of empirically-based, nuanced opportunities and threats in order to ensure a human-centered approach to generative AI in STEM education.\u003c/p\u003e \u003cp\u003eThe lack of studies involving human participants might have implications for the validity of the SWOT analysis, as the opportunities and threats are sometimes based on only a few empirical studies with human participants. This could be one reason why the SWOT analysis demonstrates seemingly contradictory results: \u003cem\u003elearning and understanding\u003c/em\u003e, \u003cem\u003eproblem solving\u003c/em\u003e and \u003cem\u003ecritical thinking\u003c/em\u003e appear both as opportunities and as threats of generative AI in STEM education. This underlines the fundamental ambivalences and tensions associated with generative AI (OECD, \u003cspan citationid=\"CR96\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e). The lack of empirical evidence on the actual implementation of generative AI in STEM education might be one reason for this divergence. Some threats and opportunities of generative AI are derived from teachers\u0026rsquo; opinions and not from an actual assessment of the impact of generative AI. Additionally, variations in the findings of different studies may result from the diverse learning scenarios being evaluated: For example, Dasari et al. (2024) not only compare a group working with AI against a traditional teaching group in mathematics but also compare an AI-only group with a group working with AI \u003cem\u003eand\u003c/em\u003e a teacher. They find that while the AI-only group shows worse mathematics results than the traditional teaching group, the AI-plus-teacher group shows better results than the traditional teaching group. Chen \u0026amp; Chang (2024) compare the impact of generative AI on digital game-based learning in physics by comparing a group learning with AI and examples (e.g., of suitable prompts) against an AI-only group, finding better learning outcomes for the former.\u003c/p\u003e \u003cp\u003eThere is a necessity for more such nuanced and differentiated research conditions to explore integration methods that enhance student learning, critical thinking, and problem-solving skills, particularly considering the role of the teachers in (successful) AI-mediated educational settings. This approach could address the seemingly contradictory results of the SWOT analysis.\u003c/p\u003e \u003cp\u003eThe focus on the research themes of \u003cem\u003eperformance\u003c/em\u003e and \u003cem\u003etool development\u003c/em\u003e \u0026ndash; while arguably problematic regarding a human-centered approach to AI \u0026ndash; could serve as a starting point for future research. The evaluation of the correctness and adequacy of AI-generated content is undeniably important for its application in the classroom (Virvou \u0026amp; Tsihrintzis, \u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). A closer look has shown that subcategories of these research themes show a considerable diversity: they range from the capacity of generative AI to solve mathematics tasks, to the generation of feedback, hints, explanations and questions, to the planning of lessons and activities as well as tutoring and student simulations. While there is some overlap among these categories (e.g., answering mathematical questions may be a prerequisite for effective tutoring), they nonetheless illustrate the breadth of the research landscape. This suggests that there is only a limited number of publications within each category, which complicates drawing definitive conclusions \u0026ndash; an issue that is exacerbated by the rapid changes of AI technologies. This also implies that the findings on strengths and weaknesses in SWOT analysis are partly based on only a few publications. Still, the strengths and weaknesses in the SWOT analysis provide initial insights into the possibilities of generative AI to enhance Human Flourishing, for example by creating personalized educational materials that meet individual student needs, and by their linguistic versatility (Kasneci et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). At the same time, biases and linguistic exclusion might risk inclusivity, equality and accessibility in STEM education.\u003c/p\u003e \u003cp\u003eFuture research should systematically increase the number of publications for the different subcategories of \u003cem\u003eperformance\u003c/em\u003e and \u003cem\u003etool development\u003c/em\u003e. An extension of research efforts within one particular subcategory could serve as a basis for an in-depth evaluation with human participants, while the evaluation of an AI system across different subcategories might allow for a more holistic picture of the educational capabilities of the system and, in a next step, of its impact on students\u0026rsquo; and teachers\u0026rsquo; flourishing.\u003c/p\u003e \u003c/div\u003e"},{"header":"7. Limitations","content":"\u003cp\u003eAlthough we have provided a broad overview of the current literature on generative AI in STEM education, including 183 full-text publications, our review only presents the state of October 2024 and does not consider research conducted afterwards. This is clearly a limitation in such a rapidly evolving research field, particularly with regard to strengths and weaknesses, which may change with newer versions of generative AI systems. Moreover, even though we have included five common databases for educational and technical research, we cannot guarantee that some relevant research results are not listed in these databases. The inclusion of preprints in the review \u0026ndash; while providing access to the latest research results \u0026ndash; presents the challenge of quality assurance, especially if preprints have been published more than a year ago and have not yet been published. However, as only 15 percent of the publications are preprints, we suppose that this issue does not impede the validity of our results too much. Regarding the SWOT analysis, it should be noted that we did not indicate the version of the generative AI system used or the date of data collection for strengths and weaknesses. It could be the case that some weaknesses are no longer weaknesses of state-of-the-art AI systems today or even vice versa, that new weaknesses emerge with more advanced tools. We therefore emphasize once again that our SWOT analysis only presents a starting point for decision-making and future research and that the exact strengths and weaknesses are dynamically changing. It could serve as a reference for future research on what aspects of generative AI systems could be interesting to focus on. Future research could also track the development of the performance of generative AI systems for STEM education over time in order to derive more nuanced conclusions about strengths and weaknesses and consequently about opportunities and threats for students and teachers. Another limitation concerning that SWOT analysis is that we only examined abstracts and conclusions for strengths, weaknesses, opportunities and threats. While we aim to achieve a more representative result by this (see section \u003cspan refid=\"Sec18\" class=\"InternalRef\"\u003e4.2.3\u003c/span\u003e), we might have missed subtleties mentioned in the publications which did not make it into the abstracts and conclusions.\u003c/p\u003e"},{"header":"8. Conclusion","content":"\u003cp\u003eEducation for Human Flourishing strives to enable students to engage in meaningful relationships and activities and enable them to lead a purposeful life for themselves and others. With rapid advances in (generative) AI technologies, the consideration of Human Flourishing and its entrenchment with AI has gained traction. This research has conducted a scoping review of the literature on generative AI in secondary STEM education to explore to what extent the current research provides a foundation for exploring Human Flourishing and AI in secondary STEM education. Following the PRISMA Extension for Scoping Reviews, we identified 183 eligible publications from five databases. These publications were examined with regard to their general characteristics, research themes and SWOTs and interpreted in the light of Human Flourishing.\u003c/p\u003e \u003cp\u003eOur findings suggest that the current literature on generative AI in STEM education does not (yet) allow for an empirically-based, holistic and inclusive exploration of Human Flourishing and AI in STEM education. Despite the necessity of broadening human capabilities due to enhanced AI technology, it was shown that more publications addressed cognitive than non-cognitive variables and research on ethical considerations is underrepresented. This complicates conclusions about the holistic impact of generative AI on students in STEM education. Regarding the need to develop new models for the future to successfully address challenges posed by generative AI, the current literature shows a lack of STEM-specific theoretical frameworks. Moreover, research is concentrated on the Global North, which increases the risk of biased solutions, potentially neglecting non-Western perspectives on Human Flourishing. Concerning the restoration of meaning to human lives in the face of generative AI, it became apparent that the focus of the current research lays on the performance of generative AI systems and on tool development, investigating primarily AI-generated content instead of human participants. This approach may hinder the adoption of a human-centered mindset towards generative AI in education and might leave the strengths of AI systems identified in the SWOT analysis unleveraged. Despite these potential shortcomings of the current research, the SWOT analysis has outlined opportunities regarding the influence of generative AI on Human Flourishing in STEM education. Threats identified in the SWOT analysis also highlighted threats to Human Flourishing in STEM education, particularly ethical issues. Moreover, we have shown that a SWOT analysis might serve as a basis for new strategies of teacher-AI-interaction in a way that preserves STEM teachers\u0026rsquo; agency, autonomy and accountability. It is based on a huge number of publications that have identified strengths and weaknesses, which offer a diverse range of empirical results. They could serve as a basis for subsequent investigations with human participants.\u003c/p\u003e \u003cp\u003eThe preliminary conclusions of this scoping review might inform future research on generative AI in STEM education in order to steer the field into a direction where Human Flourishing is at the center of educational efforts \u0026ndash; due to, in spite of, and through generative AI.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cdiv class=\"DefinitionList\"\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003ePST\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003ePre-service teacher\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eRAG\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eRetrieval augmented generation\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv class=\"DefinitionListEntry\"\u003e \u003cdiv class=\"Term\"\u003eSWOT\u003c/div\u003e \u003cdiv class=\"Description\"\u003e \u003cp\u003eStrengths, weaknesses, opportunities and threats\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFunding\u003c/h2\u003e \u003cp\u003eNot applicable as this research was not funded\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eA.F. has conducted the literature search and analysis of full-text publications. A.F. and H.S.S. have drafted an initial version of the manuscript. H.S.S. has guided the theoretical conceptualization of the article and substantially contributed to manuscript revision. Both authors have read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eA list of publications included in the scoping review as well as the coding scheme is included in the supplementary information files.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e*Abramski, K., Citraro, S., Lombardi, L., Rossetti, G., \u0026amp; Stella, M. (2023). Cognitive Network Science Reveals Bias in GPT-3, GPT-3.5 Turbo, and GPT-4 Mirroring Math Anxiety in High-School Students. \u003cem\u003eBig Data and Cognitive Computing\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(3), 124. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/bdcc7030124\u003c/span\u003e\u003cspan address=\"10.3390/bdcc7030124\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAloisi, C. (2023). The future of standardised assessment: Validity and trust in algorithms for assessment and scoring. \u003cem\u003eEuropean Journal of Education\u003c/em\u003e, \u003cem\u003e58\u003c/em\u003e(1), 98\u0026ndash;110. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/ejed.12542\u003c/span\u003e\u003cspan address=\"10.1111/ejed.12542\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmos, J. P., Ahmed Amodu, O., Azlina Raja Mahmood, R., Bolakale Abdulqudus, A., Zakaria, A. F., Rhoda Iyanda, A., Ali Bukar, U., \u0026amp; Mohd Hanapi, Z. (2025). A Bibliometric Exposition and Review on Leveraging LLMs for Programming Education. \u003cem\u003eIEEE Access\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 58364\u0026ndash;58393. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ACCESS.2025.3554627\u003c/span\u003e\u003cspan address=\"10.1109/ACCESS.2025.3554627\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Anand, A., Addala, K., Baghel, K., Goel, A., Hira, M., Gupta, R., \u0026amp; Shah, R. R. (2023). Revolutionizing High School Physics Education: A Novel Dataset. In V. Goyal, N. Kumar, S. S. Bhowmick, P. Goyal, N. Goyal, \u0026amp; D. Kumar (Eds.), \u003cem\u003eBig Data and Artificial Intelligence\u003c/em\u003e (Vol. 14418, pp. 64\u0026ndash;79). Springer Nature Switzerland. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-49601-1_5\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-49601-1_5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Anghel, E., Wang, Y., Gopalakrishnan, M., Mansukhani, P., \u0026amp; Bergner, Y. (2024). \u003cem\u003eCan LLMs evaluate items measuring collaborative problem-solving?\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.35542/osf.io/tuvks\u003c/span\u003e\u003cspan address=\"10.35542/osf.io/tuvks\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Barno, E., Albaladejo-Gonz\u0026aacute;lez, M., \u0026amp; Reich, J. (2024). Scaling Generated Feedback for Novice Teachers by Sustaining Teacher Educators\u0026rsquo; Expertise: A Design to Train LLMs with Teacher Educator Endorsement of Generated Feedback. \u003cem\u003eProceedings of the Eleventh ACM Conference on Learning @ Scale\u003c/em\u003e, 412\u0026ndash;416. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3657604.3664677\u003c/span\u003e\u003cspan address=\"10.1145/3657604.3664677\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Baytak, A. (2024). The Content Analysis of the Lesson Plans Created by ChatGPT and Google Gemini. \u003cem\u003eResearch in Social Sciences and Technology\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(1), 329\u0026ndash;350. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.46303/ressat.2024.19\u003c/span\u003e\u003cspan address=\"10.46303/ressat.2024.19\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Bellettini, C., Lodi, M., Lonati, V., Monga, M., \u0026amp; Morpurgo, A. (2023). Davinci Goes to Bebras: A Study on the Problem Solving Ability of GPT-3: \u003cem\u003eProceedings of the 15th International Conference on Computer Supported Education\u003c/em\u003e, 59\u0026ndash;69. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5220/0012007500003470\u003c/span\u003e\u003cspan address=\"10.5220/0012007500003470\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Bewersdorff, A., Se\u0026szlig;ler, K., Baur, A., Kasneci, E., \u0026amp; Nerdel, C. (2023). Assessing student errors in experimentation using artificial intelligence and large language models: A comparative study with human raters. \u003cem\u003eComputers and Education: Artificial Intelligence\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e, 100177. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2023.100177\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2023.100177\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBulathwela, S., P\u0026eacute;rez-Ortiz, M., Holloway, C., Cukurova, M., \u0026amp; Shawe-Taylor, J. (2024). Artificial Intelligence Alone Will Not Democratise Education: On Educational Inequality, Techno-Solutionism and Inclusive Tools. \u003cem\u003eSustainability\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e(2), 781. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/su16020781\u003c/span\u003e\u003cspan address=\"10.3390/su16020781\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Butgereit, L., \u0026amp; Van Staden, A. (2023). An AI Contribution to African Economic Development Through Mother Tongue Education\u0026mdash;A Linguistic Evaluation of an African-Based AI Mathematics Tutoring Bot. \u003cem\u003e2023 International Conference on Information and Communication Technology for Development for Africa (ICT4DA)\u003c/em\u003e, 7\u0026ndash;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICT4DA59526.2023.10302241\u003c/span\u003e\u003cspan address=\"10.1109/ICT4DA59526.2023.10302241\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Canonigo, A. M. (2024). Levering AI to enhance students\u0026rsquo; conceptual understanding and confidence in mathematics. \u003cem\u003eJournal of Computer Assisted Learning\u003c/em\u003e, \u003cem\u003e40\u003c/em\u003e(6), 3215\u0026ndash;3229. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/jcal.13065\u003c/span\u003e\u003cspan address=\"10.1111/jcal.13065\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Chen, C.-H., \u0026amp; Chang, C.-L. (2024). Effectiveness of AI-assisted game-based learning on science learning outcomes, intrinsic motivation, cognitive load, and learning behavior. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e29\u003c/em\u003e(14), 18621\u0026ndash;18642. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-024-12553-x\u003c/span\u003e\u003cspan address=\"10.1007/s10639-024-12553-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, X., Hu, Z., \u0026amp; Wang, C. (2024). Empowering education development through AIGC: A systematic literature review. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e29\u003c/em\u003e(13), 17485\u0026ndash;17537. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-024-12549-7\u003c/span\u003e\u003cspan address=\"10.1007/s10639-024-12549-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Cheung, K. K. C., Pun, J. K. H., \u0026amp; Li, W. (2024). Students\u0026rsquo; Holistic Reading of Socio-Scientific Texts on Climate Change in a ChatGPT Scenario. \u003cem\u003eResearch in Science Education\u003c/em\u003e, \u003cem\u003e54\u003c/em\u003e(5), 957\u0026ndash;976. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11165-024-10177-2\u003c/span\u003e\u003cspan address=\"10.1007/s11165-024-10177-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChu, R., \u0026amp; Lim, S. C. J. (2023). Education and Training for Future Engineering Teachers in the Age of Artificial Intelligence: A Bibliometric Analysis. \u003cem\u003e2023 IEEE International Conference on Industrial Engineering and Engineering\u003c/em\u003e Management \u003cem\u003e(IEEM)\u003c/em\u003e, 416\u0026ndash;420. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/IEEM58616.2023.10406630\u003c/span\u003e\u003cspan address=\"10.1109/IEEM58616.2023.10406630\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Cohn, C., Snyder, C., Montenegro, J., \u0026amp; Biswas, G. (2024). Towards A Human-in-the-Loop LLM Approach to Collaborative Discourse Analysis. \u003cem\u003eArtificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky\u003c/em\u003e, \u003cem\u003e2151\u003c/em\u003e, 11\u0026ndash;19. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-64312-5_2\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-64312-5_2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Cooper, G., \u0026amp; Tang, K.-S. (2024). Pixels and Pedagogy: Examining Science Education Imagery by Generative Artificial Intelligence. \u003cem\u003eJournal of Science Education and Technology\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(4), 556\u0026ndash;568. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10956-024-10104-0\u003c/span\u003e\u003cspan address=\"10.1007/s10956-024-10104-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCurren, R., Boniwell, I., Ryan, R. M., Oades, L., Brighouse, H., Unterhalter, E., Kristj\u0026aacute;nsson, K., De Ruyter, D., Macleod, C., Morris, I., \u0026amp; White, M. (2024). Finding consensus on well-being in education. \u003cem\u003eTheory and Research in Education\u003c/em\u003e, \u003cem\u003e22\u003c/em\u003e(2), 117\u0026ndash;157. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/14778785241259852\u003c/span\u003e\u003cspan address=\"10.1177/14778785241259852\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Daher, W., \u0026amp; Gierdien, F. (2024). Use of Language By generative AI Tools in Mathematical Problem Solving: The Case of ChatGPT. \u003cem\u003eAfrican Journal of Research in Mathematics, Science and Technology Education\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(2), 222\u0026ndash;235. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/18117295.2024.2384676\u003c/span\u003e\u003cspan address=\"10.1080/18117295.2024.2384676\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Dao, X.-Q., \u0026amp; Le, N.-B. (2023). \u003cem\u003eInvestigating the Effectiveness of ChatGPT in Mathematical Reasoning and Problem Solving: Evidence from the Vietnamese National High School Graduation Examination\u003c/em\u003e (Version 3). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2306.06331\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2306.06331\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDarvishi, A., Khosravi, H., Sadiq, S., Gašević, D., \u0026amp; Siemens, G. (2024). Impact of AI assistance on student agency. \u003cem\u003eComputers \u0026amp; Education\u003c/em\u003e, \u003cem\u003e210\u003c/em\u003e, 104967. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.compedu.2023.104967\u003c/span\u003e\u003cspan address=\"10.1016/j.compedu.2023.104967\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Dasari, D., Hendriyanto, A., Sahara, S., Suryadi, D., Muhaimin, L. H., Chao, T., \u0026amp; Fitriana, L. (2024). ChatGPT in didactical tetrahedron, does it make an exception? A case study in mathematics teaching and learning. \u003cem\u003eFrontiers in Education\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e, 1295413. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/feduc.2023.1295413\u003c/span\u003e\u003cspan address=\"10.3389/feduc.2023.1295413\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ede Ruyter, D., Oades, L. G., Waghid, Y., Ehrenfeld, J., Gilead, T., \u0026amp; Singh, N. C. (2022). Education for flourishing and flourishing in education. In A. K. Duraiappah, N. M. V. Atteveldt, G. Borst, S. Bugden, O. Ergas, T. Gilead, L. Gupta, J. Mercier, K. Pugh, S. N.C., \u0026amp; V. E. A. (Eds.), \u003cem\u003eReimagining Education: The International Science and Evidence based Education Assessment\u003c/em\u003e. UNESCO MGIEP. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.56383/XBXZ7711\u003c/span\u003e\u003cspan address=\"10.56383/XBXZ7711\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDenecke, K., Glauser, R., \u0026amp; Reichenpfader, D. (2023). Assessing the Potential and Risks of AI-Based Tools in Higher Education: Results from an eSurvey and SWOT Analysis. \u003cem\u003eTrends in Higher Education\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e(4), 667\u0026ndash;688. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/higheredu2040039\u003c/span\u003e\u003cspan address=\"10.3390/higheredu2040039\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*dos Santos, R. P. (2023). \u003cem\u003eEnhancing Chemistry Learning with ChatGPT, Bing Chat, Bard, and Claude as Agents-to-Think-With: A Comparative Case Study\u003c/em\u003e (Version 1). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://arxiv.org/abs/2311.00709\u003c/span\u003e\u003cspan address=\"https://arxiv.org/abs/2311.00709\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Drushlyak, M., Lukashova, T., Sabadosh, Y., Melnikov, I., \u0026amp; Semenikhina, O. (2024). Using ChatGPT for the Development of Critical Thinking in Youth: Example of Inequality Proof. \u003cem\u003e2024 47th MIPRO ICT and Electronics Convention (MIPRO)\u003c/em\u003e, 334\u0026ndash;339. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/MIPRO60963.2024.10569759\u003c/span\u003e\u003cspan address=\"10.1109/MIPRO60963.2024.10569759\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Due, S., Das, S., Andersen, M., L\u0026oacute;pez, B. P., Nex\u0026oslash;, S. A., \u0026amp; Clemmensen, L. (2024). \u003cem\u003eEvaluation of Large Language Models: STEM education and Gender Stereotypes\u003c/em\u003e (arXiv:2406.10133). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2406.10133\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2406.10133\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Egara, F. O., \u0026amp; Mosimege, M. (2024). Exploring the Integration of Artificial Intelligence-Based ChatGPT into Mathematics Instruction: Perceptions, Challenges, and Implications for Educators. \u003cem\u003eEducation Sciences\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(7), 742. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/educsci14070742\u003c/span\u003e\u003cspan address=\"10.3390/educsci14070742\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEnglish, L. D. (2017). Advancing Elementary and Middle School STEM Education. \u003cem\u003eInternational Journal of Science and Mathematics Education\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(S1), 5\u0026ndash;24. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10763-017-9802-x\u003c/span\u003e\u003cspan address=\"10.1007/s10763-017-9802-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Ergene, O., \u0026amp; Ergene, B. C. (2025). AI ChatBots\u0026rsquo; solutions to mathematical problems in interactive e-textbooks: Affordances and constraints from the eyes of students and teachers. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e(1), 509\u0026ndash;545. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-024-13121-z\u003c/span\u003e\u003cspan address=\"10.1007/s10639-024-13121-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eErgras, O., Gilead, T., Eilam, E., Kyza, E. E., Akar, B., Buck, M. F., \u0026amp; Kumar, S. (2022). A curricular framework for flourishing in education. In A. K. Duraiappah, N. M. V. Atteveldt, G. Borst, S. Bugden, O. Ergas, T. Gilead, L. Gupta, J. Mercier, K. Pugh, S. N.C., \u0026amp; V. E. A. (Eds.), \u003cem\u003eReimagining Education: The International Science and Evidence based Education Assessment\u003c/em\u003e. UNESCO MGIEP. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.56383/TJKT9645\u003c/span\u003e\u003cspan address=\"10.56383/TJKT9645\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFarrokhnia, M., Banihashem, S. K., Noroozi, O., \u0026amp; Wals, A. (2024). A SWOT analysis of ChatGPT: Implications for educational practice and research. \u003cem\u003eInnovations in Education and Teaching International\u003c/em\u003e, \u003cem\u003e61\u003c/em\u003e(3), 460\u0026ndash;474. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/14703297.2023.2195846\u003c/span\u003e\u003cspan address=\"10.1080/14703297.2023.2195846\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Feldman-Maggor, Y., Blonder, R., \u0026amp; Alexandron, G. (2025). Perspectives of Generative AI in Chemistry Education Within the TPACK Framework. \u003cem\u003eJournal of Science Education and Technology\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(1), 1\u0026ndash;12. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10956-024-10147-3\u003c/span\u003e\u003cspan address=\"10.1007/s10956-024-10147-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Feng, W., Lee, J., McNichols, H., Scarlatos, A., Smith, D., Woodhead, S., Ornelas, N., \u0026amp; Lan, A. (2024). Exploring Automated Distractor Generation for Math Multiple-choice Questions via Large Language Models. \u003cem\u003eFindings of the Association for Computational Linguistics: NAACL 2024\u003c/em\u003e, 3067\u0026ndash;3082. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18653/v1/2024.findings-naacl.193\u003c/span\u003e\u003cspan address=\"10.18653/v1/2024.findings-naacl.193\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Fernandez, N., Scarlatos, A., Feng, W., Woodhead, S., \u0026amp; Lan, A. (2024). \u003cem\u003eDiVERT: Distractor Generation with Variational Errors Represented as Text for Math Multiple-choice Questions\u003c/em\u003e (arXiv:2406.19356). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2406.19356\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2406.19356\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFilippi, S., \u0026amp; Motyl, B. (2024). Large Language Models (LLMs) in Engineering Education: A Systematic Review and Suggestions for Practical Adoption. \u003cem\u003eInformation\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(6), 345. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/info15060345\u003c/span\u003e\u003cspan address=\"10.3390/info15060345\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFlook, L., O\u0026rsquo;Toole, C., Ritter, J., \u0026amp; Schonert-Reichl, K. A. (2022). Human flourishing in schools. In A. K. Duraiappah, N. M. V. Atteveldt, G. Borst, S. Bugden, O. Ergas, T. Gilead, L. Gupta, J. Mercier, K. Pugh, S. N.C., \u0026amp; V. E. A. (Eds.), \u003cem\u003eReimagining Education: The International Science and Evidence based Education Assessment\u003c/em\u003e. UNESCO MGIEP. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.56383/FOQI1067\u003c/span\u003e\u003cspan address=\"10.56383/FOQI1067\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFock, A., Just, J. \u0026amp; Siller, H.-S. (2025). Opportunities and Limitations of ChatGPT in Creating STEM Tasks: A Case Study with Mathematics Pre-service Teachers. \u003cem\u003eCERME14\u003c/em\u003e, Bozen-Bolzano, Italy.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGao, Y., Zhai, X., Li, M., Lee, G., \u0026amp; Liu, X. (2025). A Multimodal Interactive Framework for Science Assessment in the Era of Generative Artificial Intelligence. \u003cem\u003eJournal of Research in Science Teaching\u003c/em\u003e, tea.70009. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/tea.70009\u003c/span\u003e\u003cspan address=\"10.1002/tea.70009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Garofalo, S. G., \u0026amp; Farenga, S. J. (2025). Science Teacher Perceptions of the State of Knowledge and Education at the Advent of Generative Artificial Intelligence Popularity. \u003cem\u003eScience \u0026amp; Education\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(2), 893\u0026ndash;912. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11191-024-00534-y\u003c/span\u003e\u003cspan address=\"10.1007/s11191-024-00534-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Gattupalli, S., Lee, W., Allessio, D., Crabtree, D., Arroyo, I., \u0026amp; Woolf, B. (2023). Exploring Pre-Service Teachers\u0026rsquo; Perceptions of Large Language Models-Generated Hints in Online Mathematics Learning. \u003cem\u003eAIED2023 Empowering Education with LLMs - the Next-Gen Interface and Content Generation\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGeiger, V., Gal, I., \u0026amp; Graven, M. (2023). The connections between citizenship education and mathematics education. \u003cem\u003eZDM \u0026ndash; Mathematics Education\u003c/em\u003e, \u003cem\u003e55\u003c/em\u003e(5), 923\u0026ndash;940. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11858-023-01521-3\u003c/span\u003e\u003cspan address=\"10.1007/s11858-023-01521-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Ghazali, S. A., Zaki, N., Ali, L., \u0026amp; Harous, S. (2024). Exploring the Potential of ChatGPT as a Substitute Teacher: A Case Study. International Journal of Information and Education Technology, 14(2), 271\u0026ndash;278. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18178/ijiet.2024.14.2.2048\u003c/span\u003e\u003cspan address=\"10.18178/ijiet.2024.14.2.2048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Gregorcic, B., \u0026amp; Pendrill, A.-M. (2023). ChatGPT and the frustrated Socrates. \u003cem\u003ePhysics Education\u003c/em\u003e, \u003cem\u003e58\u003c/em\u003e(3), 035021. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1088/1361-6552/acc299\u003c/span\u003e\u003cspan address=\"10.1088/1361-6552/acc299\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Guler, N. K., Dertli, Z. G., Boran, E., \u0026amp; Yildiz, B. (2024). An artificial intelligence application in mathematics education: Evaluating ChatGPT\u0026rsquo;s academic achievement in a mathematics exam. \u003cem\u003ePedagogical Research\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(2), em0188. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.29333/pr/14145\u003c/span\u003e\u003cspan address=\"10.29333/pr/14145\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Han, Z. F., Lin, J., Gurung, A., Thomas, D. R., Chen, E., Borchers, C., Gupta, S., \u0026amp; Koedinger, K. R. (2024). \u003cem\u003eImproving Assessment of Tutoring Practices using Retrieval-Augmented Generation\u003c/em\u003e (Version 1). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2402.14594\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2402.14594\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHarris, H., \u0026amp; Kittur, J. (2024). Generative Artificial Intelligence in Undergraduate Engineering: A Systematic Literature Review. \u003cem\u003e2024 ASEE Annual Conference \u0026amp; Exposition Proceedings\u003c/em\u003e, 47492. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18260/1-2--47492\u003c/span\u003e\u003cspan address=\"10.18260/1-2--47492\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Hashem, R., Ali, N., El Zein, F., Fidalgo, P., \u0026amp; Abu Khurma, O. (2024). AI to the rescue: Exploring the potential of ChatGPT as a teacher ally for workload relief and burnout prevention. \u003cem\u003eResearch and Practice in Technology Enhanced Learning\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e, 023. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.58459/rptel.2024.19023\u003c/span\u003e\u003cspan address=\"10.58459/rptel.2024.19023\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHelms, M. M., \u0026amp; Nixon, J. (2010). Exploring SWOT analysis \u0026ndash; where are we now?: A review of academic research from the last decade. \u003cem\u003eJournal of Strategy and Management\u003c/em\u003e, \u003cem\u003e3\u003c/em\u003e(3), 215\u0026ndash;251. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1108/17554251011064837\u003c/span\u003e\u003cspan address=\"10.1108/17554251011064837\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Henkel, O., Horne-Robinson, H., Dyshel, M., Ch, N., Moreau-Pernet, B., \u0026amp; Abood, R. (2024). \u003cem\u003eLearning to Love Edge Cases in Formative Math Assessment: Using the AMMORE Dataset and Chain-of-Thought Prompting to Improve Grading Accuracy\u003c/em\u003e (arXiv:2409.17904). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2409.17904\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2409.17904\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Henkel, O., Horne-Robinson, H., Kozhakhmetova, N., \u0026amp; Lee, A. (2024). Effective and Scalable Math Support: Experimental Evidence on the Impact of an AI-Math Tutor in Ghana. In A. M. Olney, I.-A. Chounta, Z. Liu, O. C. Santos, \u0026amp; I. I. Bittencourt (Eds.), \u003cem\u003eArtificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky\u003c/em\u003e (Vol. 2150, pp. 373\u0026ndash;381). Springer Nature Switzerland. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-64315-6_34\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-64315-6_34\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolstein, K., \u0026amp; Aleven, V. (2022). Designing for human\u0026ndash;AI complementarity in K-12 education. \u003cem\u003eAI Magazine\u003c/em\u003e, \u003cem\u003e43\u003c/em\u003e(2), 239\u0026ndash;248. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/aaai.12058\u003c/span\u003e\u003cspan address=\"10.1002/aaai.12058\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Hu, B., Zheng, L., Zhu, J., Ding, L., Wang, Y., \u0026amp; Gu, X. (2024). Teaching Plan Generation and Evaluation With GPT-4: Unleashing the Potential of LLM in Instructional Design. \u003cem\u003eIEEE Transactions on Learning Technologies\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e, 1445\u0026ndash;1459. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/TLT.2024.3384765\u003c/span\u003e\u003cspan address=\"10.1109/TLT.2024.3384765\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHumble, N. (2024). Risk management strategy for generative AI in computing education: How to handle the strengths, weaknesses, opportunities, and threats? \u003cem\u003eInternational Journal of Educational Technology in Higher Education\u003c/em\u003e, \u003cem\u003e21\u003c/em\u003e(1), 61. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s41239-024-00494-x\u003c/span\u003e\u003cspan address=\"10.1186/s41239-024-00494-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Jia, J., Wang, T., Zhang, Y., \u0026amp; Wang, G. (2024). The comparison of general tips for mathematical problem solving generated by generative AI with those generated by human teachers. \u003cem\u003eAsia Pacific Journal of Education\u003c/em\u003e, \u003cem\u003e44\u003c/em\u003e(1), 8\u0026ndash;28. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/02188791.2023.2286920\u003c/span\u003e\u003cspan address=\"10.1080/02188791.2023.2286920\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Jin, H., Yoo, M., Park, J., Lee, Y., Wang, X., \u0026amp; Kim, J. (2024). \u003cem\u003eTeachTune: Reviewing Pedagogical Agents Against Diverse Student Profiles with Simulated Students\u003c/em\u003e (Version 3). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2410.04078\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2410.04078\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Kakarla, S., Thomas, D., Lin, J., Gupta, S., \u0026amp; Koedinger, K. R. (2024). \u003cem\u003eUsing Large Language Models to Assess Tutors\u0026rsquo; Performance in Reacting to Students Making Math Errors\u003c/em\u003e (arXiv:2401.03238). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2401.03238\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2401.03238\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKarakuş, N., Gedik, K., \u0026amp; Kazazoğlu, S. (2025). Ethical Decision-Making in Education: A Comparative Study of Teachers and Artificial Intelligence in Ethical Dilemmas. \u003cem\u003eBehavioral Sciences\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(4), 469. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/bs15040469\u003c/span\u003e\u003cspan address=\"10.3390/bs15040469\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKasneci, E., Sessler, K., K\u0026uuml;chemann, S., Bannert, M., Dementieva, D., Fischer, F., Gasser, U., Groh, G., G\u0026uuml;nnemann, S., H\u0026uuml;llermeier, E., Krusche, S., Kutyniok, G., Michaeli, T., Nerdel, C., Pfeffer, J., Poquet, O., Sailer, M., Schmidt, A., Seidel, T., \u0026hellip; Kasneci, G. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. \u003cem\u003eLearning and Individual Differences\u003c/em\u003e, \u003cem\u003e103\u003c/em\u003e, 102274. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.lindif.2023.102274\u003c/span\u003e\u003cspan address=\"10.1016/j.lindif.2023.102274\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKong, S.-C., \u0026amp; Yang, Y. (2024). A Human-Centered Learning and Teaching Framework Using Generative Artificial Intelligence for Self-Regulated Learning Development Through Domain Knowledge Learning in K\u0026ndash;12 Settings. \u003cem\u003eIEEE Transactions on Learning Technologies\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e, 1562\u0026ndash;1573. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/TLT.2024.3392830\u003c/span\u003e\u003cspan address=\"10.1109/TLT.2024.3392830\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKorseberg, L., \u0026amp; Elken, M. (2025). Waiting for the revolution: How higher education institutions initially responded to ChatGPT. \u003cem\u003eHigher Education\u003c/em\u003e, \u003cem\u003e89\u003c/em\u003e(4), 953\u0026ndash;968. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10734-024-01256-4\u003c/span\u003e\u003cspan address=\"10.1007/s10734-024-01256-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*K\u0026uuml;chemann, S., Steinert, S., Revenga, N., Schweinberger, M., Dinc, Y., Avila, K. E., \u0026amp; Kuhn, J. (2023). Can ChatGPT support prospective teachers in physics task development? \u003cem\u003ePhysical Review Physics Education Research\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(2), 020128. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1103/PhysRevPhysEducRes.19.020128\u003c/span\u003e\u003cspan address=\"10.1103/PhysRevPhysEducRes.19.020128\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuckartz, U. (2019). Qualitative Text Analysis: A Systematic Approach. In G. Kaiser \u0026amp; N. Presmeg (Eds.), \u003cem\u003eCompendium for Early Career Researchers in Mathematics Education\u003c/em\u003e (pp. 181\u0026ndash;197). Springer International Publishing. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-030-15636-7_8\u003c/span\u003e\u003cspan address=\"10.1007/978-3-030-15636-7_8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuckartz, U., \u0026amp; R\u0026auml;diker, S. (2023). \u003cem\u003eQualitative content analysis: Methods, practice and software\u003c/em\u003e (2nd edition). SAGE.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Kunz, S., \u0026amp; Steffen, A. (2024). Criticize my Code, not me: Using AI-generated Feedback in Computer Science Teaching. \u003cem\u003eInternational Conference on Gender Research\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(1), 200\u0026ndash;207. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.34190/icgr.7.1.2002\u003c/span\u003e\u003cspan address=\"10.34190/icgr.7.1.2002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Latif, E., \u0026amp; Zhai, X. (2024). Fine-tuning ChatGPT for automatic scoring. \u003cem\u003eComputers and Education: Artificial Intelligence\u003c/em\u003e, \u003cem\u003e6\u003c/em\u003e, 100210. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2024.100210\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2024.100210\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Lee, G.-G., Latif, E., Wu, X., Liu, N., \u0026amp; Zhai, X. (2024). Applying large language models and chain-of-thought for automatic scoring. \u003cem\u003eComputers and Education: Artificial Intelligence\u003c/em\u003e, \u003cem\u003e6\u003c/em\u003e, 100213. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.caeai.2024.100213\u003c/span\u003e\u003cspan address=\"10.1016/j.caeai.2024.100213\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Levonian, Z., Li, C., Zhu, W., Gade, A., Henkel, O., Postle, M.-E., \u0026amp; Xing, W. (2023). \u003cem\u003eRetrieval-augmented Generation to Improve Math Question-Answering: Trade-offs Between Groundedness and Human Preference\u003c/em\u003e (arXiv:2310.03184). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2310.03184\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2310.03184\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Li, R., Wang, Y., Zheng, C., Jiang, Y.-H., \u0026amp; Jiang, B. (2024). Generating Contextualized Mathematics Multiple-Choice Questions Utilizing Large Language Models. In A. M. Olney, I.-A. Chounta, Z. Liu, O. C. Santos, \u0026amp; I. I. Bittencourt (Eds.), \u003cem\u003eArtificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky\u003c/em\u003e (Vol. 2150, pp. 494\u0026ndash;501). Springer Nature Switzerland. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-64315-6_48\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-64315-6_48\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Li, T., Ji, Y., \u0026amp; Zhan, Z. (2024). Expert or machine? Comparing the effect of pairing student teacher with in-service teacher and ChatGPT on their critical thinking, learning performance, and cognitive load in an integrated-STEM course. \u003cem\u003eAsia Pacific Journal of Education\u003c/em\u003e, \u003cem\u003e44\u003c/em\u003e(1), 45\u0026ndash;60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/02188791.2024.2305163\u003c/span\u003e\u003cspan address=\"10.1080/02188791.2024.2305163\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Li, Z., \u0026amp; Ironsi, C. S. (2024). The Efficacy of Generative Artificial Intelligence in Developing Science Education Preservice Teachers\u0026rsquo; Writing Skills: An Experimental Approach. \u003cem\u003eJournal of Science Education and Technology\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10956-024-10148-2\u003c/span\u003e\u003cspan address=\"10.1007/s10956-024-10148-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Li, Z.-Z., Zhang, M.-L., Yin, F., Ji, Z.-L., Bai, J.-F., Pan, Z.-R., Zeng, F.-H., Xu, J., Zhang, J.-X., \u0026amp; Liu, C.-L. (2024). \u003cem\u003eCMMaTH: A Chinese Multi-modal Math Skill Evaluation Benchmark for Foundation Models\u003c/em\u003e (arXiv:2407.12023). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2407.12023\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2407.12023\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Lin, T., \u0026amp; Riccomini, P. (2024). Inclusive Education in Rural Settings: Leveraging Technology for Improved Math Learning Outcomes Among Students With Disabilities. \u003cem\u003eJournal of Special Education Technology\u003c/em\u003e, \u003cem\u003e40\u003c/em\u003e(2), 219\u0026ndash;226. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/01626434241263045\u003c/span\u003e\u003cspan address=\"10.1177/01626434241263045\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eListyaningrum, P., Retnawati, H., Harun, H., \u0026amp; Ibda, H. (2024). Digital learning using ChatGPT in elementary school mathematics learning: A systematic literature review. \u003cem\u003eIndonesian Journal of Electrical Engineering and Computer Science\u003c/em\u003e, \u003cem\u003e36\u003c/em\u003e(3), 1701. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.11591/ijeecs.v36.i3.pp1701-1710\u003c/span\u003e\u003cspan address=\"10.11591/ijeecs.v36.i3.pp1701-1710\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Liu, N., Sonkar, S., Le, M., \u0026amp; Baraniuk, R. (2024). \u003cem\u003eMalAlgoQA: Pedagogical Evaluation of Counterfactual Reasoning in Large Language Models and Implications for AI in Education\u003c/em\u003e (arXiv:2407.00938). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2407.00938\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2407.00938\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLoke, D., Radloff, J., \u0026amp; Yeter, I. H. (2024). Board 165: K-12 STEM Teachers\u0026rsquo; Perceptions of Artificial Intelligence: A PRISMA-tic Approach (Work-in-Progress). \u003cem\u003e2024 ASEE Annual Conference \u0026amp; Exposition Proceedings\u003c/em\u003e, 46727. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18260/1-2--46727\u003c/span\u003e\u003cspan address=\"10.18260/1-2--46727\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Long, Y., Luo, H., \u0026amp; Zhang, Y. (2024). Evaluating large language models in analysing classroom dialogue. \u003cem\u003eNpj Science of Learning\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(1), 60. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41539-024-00273-3\u003c/span\u003e\u003cspan address=\"10.1038/s41539-024-00273-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Lu, J., Zheng, R., Gong, Z., \u0026amp; Xu, H. (2024). Supporting Teachers\u0026rsquo; Professional Development With Generative AI: The Effects on Higher Order Thinking and Self-Efficacy. \u003cem\u003eIEEE Transactions on Learning Technologies\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e, 1267\u0026ndash;1277. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/TLT.2024.3369690\u003c/span\u003e\u003cspan address=\"10.1109/TLT.2024.3369690\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMayring, P. (2014). \u003cem\u003eQualitative content analysis: Theoretical foundation, basic procedures and software solution\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://nbn-resolving.de/urn:nbn:de:0168-ssoar-395173\u003c/span\u003e\u003cspan address=\"http://nbn-resolving.de/urn:nbn:de:0168-ssoar-395173\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcGowan, J., Sampson, M., Salzwedel, D. M., Cogo, E., Foerster, V., \u0026amp; Lefebvre, C. (2016). PRESS Peer Review of Electronic Search Strategies: 2015 Guideline Statement. \u003cem\u003eJournal of Clinical Epidemiology\u003c/em\u003e, \u003cem\u003e75\u003c/em\u003e, 40\u0026ndash;46. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jclinepi.2016.01.021\u003c/span\u003e\u003cspan address=\"10.1016/j.jclinepi.2016.01.021\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*McNichols, H., Feng, W., Lee, J., Scarlatos, A., Smith, D., Woodhead, S., \u0026amp; Lan, A. (2023). \u003cem\u003eAutomated Distractor and Feedback Generation for Math Multiple-choice Questions via In-context Learning\u003c/em\u003e (arXiv:2308.03234). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2308.03234\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2308.03234\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*McNichols, H., Lee, J., Fancsali, S., Ritter, S., \u0026amp; Lan, A. (2024). \u003cem\u003eCan Large Language Models Replicate ITS Feedback on Open-Ended Math Questions?\u003c/em\u003e (arXiv:2405.06414). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2405.06414\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2405.06414\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMerzifonluoglu, A., \u0026amp; Gunes, H. (2025). Shifting Dynamics: Who Holds the Reins in Decision-Making With Artificial Intelligence Tools? Perspectives of Gen Z Pre‐Service Teachers. \u003cem\u003eEuropean Journal of Education\u003c/em\u003e, \u003cem\u003e60\u003c/em\u003e(1), e70053. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/ejed.70053\u003c/span\u003e\u003cspan address=\"10.1111/ejed.70053\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiao, F., \u0026amp; Cukurova, M. (2024). \u003cem\u003eAI competency framework for teachers\u003c/em\u003e. UNESCO. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.54675/ZJTE2084\u003c/span\u003e\u003cspan address=\"10.54675/ZJTE2084\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMiao, F., \u0026amp; Holmes, W. (2023). \u003cem\u003eGuidance for generative AI in education and research\u003c/em\u003e. UNESCO. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.54675/EWZM9535\u003c/span\u003e\u003cspan address=\"10.54675/EWZM9535\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMishra, P., Warr, M., \u0026amp; Islam, R. (2023). TPACK in the age of ChatGPT and Generative AI. \u003cem\u003eJournal of Digital Learning in Teacher Education\u003c/em\u003e, \u003cem\u003e39\u003c/em\u003e(4), 235\u0026ndash;251. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/21532974.2023.2247480\u003c/span\u003e\u003cspan address=\"10.1080/21532974.2023.2247480\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Moreau-Pernet, B., Tian, Y., Sawaya, S., Foltz, P., Cao, J., Milne, B., \u0026amp; Christie, T. (2024). Classifying Tutor Discursive Moves at Scale in Mathematics Classrooms with Large Language Models. \u003cem\u003eProceedings of the Eleventh ACM Conference on Learning @ Scale\u003c/em\u003e, 361\u0026ndash;365. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3657604.3664664\u003c/span\u003e\u003cspan address=\"10.1145/3657604.3664664\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMunn, Z., Peters, M. D. J., Stern, C., Tufanaru, C., McArthur, A., \u0026amp; Aromataris, E. (2018). Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach. \u003cem\u003eBMC Medical Research Methodology\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(1), 143. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12874-018-0611-x\u003c/span\u003e\u003cspan address=\"10.1186/s12874-018-0611-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNavigli, R., Conia, S., \u0026amp; Ross, B. (2023). Biases in Large Language Models: Origins, Inventory, and Discussion. \u003cem\u003eJournal of Data and Information Quality\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(2), 1\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3597307\u003c/span\u003e\u003cspan address=\"10.1145/3597307\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNg, D. T. K., Leung, J. K. L., Su, J., Ng, R. C. W., \u0026amp; Chu, S. K. W. (2023). Teachers\u0026rsquo; AI digital competencies and twenty-first century skills in the post-pandemic world. \u003cem\u003eEducational Technology Research and Development\u003c/em\u003e, \u003cem\u003e71\u003c/em\u003e(1), 137\u0026ndash;161. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11423-023-10203-6\u003c/span\u003e\u003cspan address=\"10.1007/s11423-023-10203-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Nicula, B., Dascalu, M., Arner, T., Balyan, R., \u0026amp; McNamara, D. S. (2023). Automated Assessment of Comprehension Strategies from Self-Explanations Using LLMs. \u003cem\u003eInformation\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(10), 567. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/info14100567\u003c/span\u003e\u003cspan address=\"10.3390/info14100567\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Norberg, K. A., Almoubayyed, H., De Ley, L., Murphy, A., Weldon, K., \u0026amp; Ritter, S. (2024). Rewriting Content with GPT-4 to Support Emerging Readers in Adaptive Mathematics Software. \u003cem\u003eInternational Journal of Artificial Intelligence in Education\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40593-024-00420-2\u003c/span\u003e\u003cspan address=\"10.1007/s40593-024-00420-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOECD. (2023). \u003cem\u003eThe New Flourishing Agenda in Education: A Report on the Current Theoretical State of Play\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.oecd.org/content/dam/oecd/en/about/projects/edu/hpst/EHF_The%20new%20flourishing%20agenda_Kristjansson.pdf\u003c/span\u003e\u003cspan address=\"https://www.oecd.org/content/dam/oecd/en/about/projects/edu/hpst/EHF_The%20new%20flourishing%20agenda_Kristjansson.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOECD. (2024a). \u003cem\u003ePISA 2022 Results (Volume V): Learning Strategies and Attitudes for Life\u003c/em\u003e. OECD. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1787/c2e44201-en\u003c/span\u003e\u003cspan address=\"10.1787/c2e44201-en\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOECD. (2024b). \u003cem\u003ePISA High Performing Systems for Tomorrow: Education for Human Flourishing\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://one.oecd.org/document/EDU/PISA/GB(2024)10/en/pdf\u003c/span\u003e\u003cspan address=\"https://one.oecd.org/document/EDU/PISA/GB(2024)10/en/pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOpesemowo, O. A. G., \u0026amp; Adewuyi, H. O. (2024). A systematic review of artificial intelligence in mathematics education: The emergence of 4IR. \u003cem\u003eEurasia Journal of Mathematics, Science and Technology Education\u003c/em\u003e, \u003cem\u003e20\u003c/em\u003e(7), em2478. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.29333/ejmste/14762\u003c/span\u003e\u003cspan address=\"10.29333/ejmste/14762\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Parra, V., Sureda, P., Corica, A., Schiaffino, S., \u0026amp; Godoy, D. (2024). Can Generative AI Solve Geometry Problems? Strengths and Weaknesses of LLMs for Geometric Reasoning in Spanish. \u003cem\u003eInternational Journal of Interactive Multimedia and Artificial Intelligence\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(5), 65. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.9781/ijimai.2024.02.009\u003c/span\u003e\u003cspan address=\"10.9781/ijimai.2024.02.009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Pavlova, N. H. (2024). Flipped dialogic learning method with ChatGPT: A case study. \u003cem\u003eInternational Electronic Journal of Mathematics Education\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(1), em0764. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.29333/iejme/14025\u003c/span\u003e\u003cspan address=\"10.29333/iejme/14025\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePendleton, D. (1975). Calculators in the classroom. \u003cem\u003eScience News\u003c/em\u003e, \u003cem\u003e107\u003c/em\u003e, 175\u0026ndash;181.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePepin, B., Buchholtz, N., \u0026amp; Salinas-Hern\u0026aacute;ndez, U. (2025). A Scoping Survey of ChatGPT in Mathematics Education. \u003cem\u003eDigital Experiences in Mathematics Education\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(1), 9\u0026ndash;41. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40751-025-00172-1\u003c/span\u003e\u003cspan address=\"10.1007/s40751-025-00172-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Prihar, E., Lee, M., Hopman, M., Kalai, A. T., Vempala, S., Wang, A., Wickline, G., Murray, A., \u0026amp; Heffernan, N. (2023). Comparing Different Approaches to Generating Mathematics Explanations Using Large Language Models. In N. Wang, G. Rebolledo-Mendez, V. Dimitrova, N. Matsuda, \u0026amp; O. C. Santos (Eds.), \u003cem\u003eArtificial Intelligence in Education. Posters and Late Breaking Results, Workshops and Tutorials, Industry and Innovation Tracks, Practitioners, Doctoral Consortium and Blue Sky\u003c/em\u003e (Vol. 1831, pp. 290\u0026ndash;295). Springer Nature Switzerland. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-36336-8_45\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-36336-8_45\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRaihan, N., Siddiq, M. L., Santos, J. C. S., \u0026amp; Zampieri, M. (2024). \u003cem\u003eLarge Language Models in Computer Science Education: A Systematic Literature Review\u003c/em\u003e (arXiv:2410.16349). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/arXiv.2410.16349\u003c/span\u003e\u003cspan address=\"10.48550/arXiv.2410.16349\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRethlefsen, M. L., Kirtley, S., Waffenschmidt, S., Ayala, A. P., Moher, D., Page, M. J., Koffel, J. B., PRISMA-S Group, Blunt, H., Brigham, T., Chang, S., Clark, J., Conway, A., Couban, R., De Kock, S., Farrah, K., Fehrmann, P., Foster, M., Fowler, S. A., \u0026hellip; Young, S. (2021). PRISMA-S: An extension to the PRISMA Statement for Reporting Literature Searches in Systematic Reviews. \u003cem\u003eSystematic Reviews\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(1), 39. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13643-020-01542-z\u003c/span\u003e\u003cspan address=\"10.1186/s13643-020-01542-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Ribeiro, A. R. A., Rosotti Navarro, E., \u0026amp; Kalinke, M. A. (2024). O uso do ChatGPT para resolver problemas matem\u0026aacute;ticos sobre grandezas direta e inversamente proporcionais. \u003cem\u003eRevista Pesquisa Qualitativa\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(30), 01\u0026ndash;21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.33361/RPQ.2024.v.12.n.30.716\u003c/span\u003e\u003cspan address=\"10.33361/RPQ.2024.v.12.n.30.716\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoberts, J., \u0026amp; Mohamed, A. (2024). Generative AI in CS Education: Literature Review through a SWOT Lens. \u003cem\u003eThe 26th Western Canadian Conference on Computing Education\u003c/em\u003e, 1\u0026ndash;6. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3660650.3660657\u003c/span\u003e\u003cspan address=\"10.1145/3660650.3660657\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Rouzegar, H., \u0026amp; Makrehchi, M. (2024). \u003cem\u003eGenerative AI for Enhancing Active Learning in Education: A Comparative Study of GPT-3.5 and GPT-4 in Crafting Customized Test Questions\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2406.13903\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2406.13903\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRussell, S., \u0026amp; Norvig, P. (2021). \u003cem\u003eArtificial Intelligence: A Modern Approach\u003c/em\u003e (4th ed.). Pearson. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://elibrary.pearson.de/book/99.150005/9781292401171\u003c/span\u003e\u003cspan address=\"https://elibrary.pearson.de/book/99.150005/9781292401171\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchinkel, A., Wolbert, L., Pedersen, J. B. W., \u0026amp; De Ruyter, D. J. (2023). Human Flourishing, Wonder, and Education. \u003cem\u003eStudies in Philosophy and Education\u003c/em\u003e, \u003cem\u003e42\u003c/em\u003e(2), 143\u0026ndash;162. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11217-022-09851-7\u003c/span\u003e\u003cspan address=\"10.1007/s11217-022-09851-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Schorcht, S., Baumanns, L., Buchholtz, N., Huget, J., Peters, F., \u0026amp; Pohl, M. (2024). The tasty world of irrational numbers \u0026ndash; explorative mathematical visualizations generated by AI image models. \u003cem\u003eJournal of Mathematics and the Arts\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(3\u0026ndash;4), 272\u0026ndash;289. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/17513472.2024.2407273\u003c/span\u003e\u003cspan address=\"10.1080/17513472.2024.2407273\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Schorcht, S., Buchholtz, N., \u0026amp; Baumanns, L. (2024). Prompt the problem \u0026ndash; investigating the mathematics educational quality of AI-supported problem solving by comparing prompt techniques. \u003cem\u003eFrontiers in Education\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e, 1386075. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/feduc.2024.1386075\u003c/span\u003e\u003cspan address=\"10.3389/feduc.2024.1386075\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Shankar, S. K., Pothancheri, G., Sasi, D., \u0026amp; Mishra, S. (2025). Bringing Teachers in the Loop: Exploring Perspectives on Integrating Generative AI in Technology-Enhanced Learning. \u003cem\u003eInternational Journal of Artificial Intelligence in Education\u003c/em\u003e, \u003cem\u003e35\u003c/em\u003e(1), 155\u0026ndash;180. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40593-024-00428-8\u003c/span\u003e\u003cspan address=\"10.1007/s40593-024-00428-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSiller, H.-S., Fock, A., \u0026amp; Just, J. (2025). \u003cem\u003eMathematics, Sustainability and Artificial Intelligence \u0026ndash; Reflecting the Challenges, Potentials \u0026amp; Opportunities of the Socio-ecological in Mathematics Lessons\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSiller, H.-S., G\u0026uuml;nster, S. M., \u0026amp; Geiger, V. (2024). Mathematics as a Central Focus in STEM \u0026ndash; Theoretical and Practical Insights from a Special Study Program Within Pre-service (Prospective) Teacher Education. In Y. Li, Z. Zeng, \u0026amp; N. Song (Eds.), \u003cem\u003eDisciplinary and Interdisciplinary Education in STEM\u003c/em\u003e (pp. 317\u0026ndash;343). Springer Nature Switzerland. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-3-031-52924-5_15\u003c/span\u003e\u003cspan address=\"10.1007/978-3-031-52924-5_15\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Singh, A., Ehtesham, A., Kumar, S., Gupta, G. K., \u0026amp; Khoei, T. T. (2024). Encouraging Responsible Use of Generative AI in Education: A Reward-Based Learning Approach. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://arxiv.org/abs/2407.15022v1\u003c/span\u003e\u003cspan address=\"https://arxiv.org/abs/2407.15022v1\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e* Sonkar, S., Chen, X., Le, M., Liu, N [Naiming], Basu Mallick, D., \u0026amp; Baraniuk, R [Richard] (2024). Code Soliloquies for Accurate Calculations in Large Language Models. In Proceedings of the 14th Learning Analytics and Knowledge Conference (pp. 828\u0026ndash;835). ACM. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3636555.3636889\u003c/span\u003e\u003cspan address=\"10.1145/3636555.3636889\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Spreitzer, C., Straser, O., Zehetmeier, S., \u0026amp; Maa\u0026szlig;, K. (2024). Mathematical Modelling Abilities of Artificial Intelligence Tools: The Case of ChatGPT. \u003cem\u003eEducation Sciences\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(7), 698. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/educsci14070698\u003c/span\u003e\u003cspan address=\"10.3390/educsci14070698\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStevenson, M. (2022). \u003cem\u003eEducation for Human Flourishing\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSu, F., \u0026amp; Jackson, C. (2020). \u003cem\u003eMathematics for human flourishing\u003c/em\u003e. Yale University Press.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Taani, O., \u0026amp; Alabidi, S. (2024). ChatGPT in education: Benefits and challenges of ChatGPT for mathematics and science teaching practices. \u003cem\u003eInternational Journal of Mathematical Education in Science and Technology\u003c/em\u003e, 1\u0026ndash;30. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/0020739X.2024.2357341\u003c/span\u003e\u003cspan address=\"10.1080/0020739X.2024.2357341\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTaulli, T. (2023). \u003cem\u003eGenerative AI: How ChatGPT and Other AI Tools Will Revolutionize Business\u003c/em\u003e. Apress. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/978-1-4842-9367-6\u003c/span\u003e\u003cspan address=\"10.1007/978-1-4842-9367-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Teegavarapu, R. R., \u0026amp; Sanghvi, H. (2023). Analyzing the Competitive Mathematical Problem-Solving Skills of ChatGPT. \u003cem\u003e2023 International Conference on Data Science, Agents \u0026amp; Artificial Intelligence (ICDSAAI)\u003c/em\u003e, 1\u0026ndash;7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/ICDSAAI59313.2023.10452659\u003c/span\u003e\u003cspan address=\"10.1109/ICDSAAI59313.2023.10452659\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTricco, A. C., Lillie, E., Zarin, W., O\u0026rsquo;Brien, K. K., Colquhoun, H., Levac, D., Moher, D., Peters, M. D. J., Horsley, T., Weeks, L., Hempel, S., Akl, E. A., Chang, C., McGowan, J., Stewart, L., Hartling, L., Aldcroft, A., Wilson, M. G., Garritty, C., \u0026hellip; Straus, S. E. (2018). PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. \u003cem\u003eAnnals of Internal Medicine\u003c/em\u003e, \u003cem\u003e169\u003c/em\u003e(7), 467\u0026ndash;473. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.7326/M18-0850\u003c/span\u003e\u003cspan address=\"10.7326/M18-0850\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*van Pham, P. L., Duc, A. V., Hoang, N. M., Do, X. L., \u0026amp; Luu, A. T. (2024). ChatGPT as a Math Questioner? Evaluating ChatGPT on Generating Pre-university Math Questions. In J. Hong, J. W. Park, \u0026amp; A. Przybyłek (Eds.), Proceedings of the 39th ACM/SIGAPP Symposium on Applied Computing (pp. 65\u0026ndash;73). ACM. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1145/3605098.3636030\u003c/span\u003e\u003cspan address=\"10.1145/3605098.3636030\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Vank\u0026uacute;š, P. (2024). Generative Artificial Intelligence on Mobile Devices in the University Preparation of Future Teachers of Mathematics. \u003cem\u003eInternational Journal of Interactive Mobile Technologies (iJIM)\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(18), 19\u0026ndash;33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3991/ijim.v18i18.51221\u003c/span\u003e\u003cspan address=\"10.3991/ijim.v18i18.51221\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Vasconcelos, M. A. R., \u0026amp; Dos Santos, R. P. (2023). Enhancing STEM learning with ChatGPT and Bing Chat as objects to think with: A case study. \u003cem\u003eEurasia Journal of Mathematics, Science and Technology Education\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(7), em2296. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.29333/ejmste/13313\u003c/span\u003e\u003cspan address=\"10.29333/ejmste/13313\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVirvou, M., \u0026amp; Tsihrintzis, G. A. (2024). Artificial Intelligence Chatbots and Social Robots in Education: FEPER Framework for Efficiency, Pedagogical and Ethical Requirements. \u003cem\u003e2024 15th International Conference on Information, Intelligence, Systems \u0026amp; Applications (IISA)\u003c/em\u003e, 1\u0026ndash;8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/IISA62523.2024.10786710\u003c/span\u003e\u003cspan address=\"10.1109/IISA62523.2024.10786710\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVon Davier, M., Kennedy, A., Reynolds, K., Fishbein, B., Khorramdel, L., Aldrich, C., Bookbinder, A., Bezirhan, U., \u0026amp; Yin, L. (2024). \u003cem\u003eTIMSS 2023 International Results in Mathematics and Science\u003c/em\u003e. TIMSS \u0026amp; PIRLS International Study Center, Boston College. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.6017/lse.tpisc.timss.rs6460\u003c/span\u003e\u003cspan address=\"10.6017/lse.tpisc.timss.rs6460\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Wardat, Y., Tashtoush, M. A., AlAli, R., \u0026amp; Jarrah, A. M. (2023). ChatGPT: A revolutionary tool for teaching and learning mathematics. \u003cem\u003eEurasia Journal of Mathematics, Science and Technology Education\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(7), em2286. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.29333/ejmste/13272\u003c/span\u003e\u003cspan address=\"10.29333/ejmste/13272\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Wei, X. (2024). Evaluating chatGPT-4 and chatGPT-4o: Performance insights from NAEP mathematics problem solving. \u003cem\u003eFrontiers in Education\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e, 1452570. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/feduc.2024.1452570\u003c/span\u003e\u003cspan address=\"10.3389/feduc.2024.1452570\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilloughby, S. (1985). Curriculum trends: Mathematics: Will calculators rot our minds? \u003cem\u003eEducational Leadership\u003c/em\u003e, \u003cem\u003e43\u003c/em\u003e, 90\u0026ndash;91.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWolbert, L. S., De Ruyter, D. J., \u0026amp; Schinkel, A. (2015). Formal criteria for the concept of human flourishing: The first step in defending flourishing as an ideal aim of education. \u003cem\u003eEthics and Education\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(1), 118\u0026ndash;129. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/17449642.2014.998032\u003c/span\u003e\u003cspan address=\"10.1080/17449642.2014.998032\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, W., \u0026amp; Ouyang, F. (2022). A systematic review of AI role in the educational system based on a proposed conceptual framework. \u003cem\u003eEducation and Information Technologies\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e(3), 4195\u0026ndash;4223. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s10639-021-10774-y\u003c/span\u003e\u003cspan address=\"10.1007/s10639-021-10774-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYacoubian, H. A. (2018). Scientific literacy for democratic decision-making. \u003cem\u003eInternational Journal of Science Education\u003c/em\u003e, \u003cem\u003e40\u003c/em\u003e(3), 308\u0026ndash;327. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/09500693.2017.1420266\u003c/span\u003e\u003cspan address=\"10.1080/09500693.2017.1420266\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Yan, Y., Wang, S., Huo, J., Li, H., Li, B., Su, J., Gao, X., Zhang, Y.-F., Xu, T., Chu, Z., Zhong, A., Wang, K., Xiong, H., Yu, P. S., Hu, X., \u0026amp; Wen, Q. (2024). \u003cem\u003eErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection\u003c/em\u003e (Version 2). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2410.04509\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2410.04509\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang, Y., Chen, L., He, W., Sun, D., \u0026amp; Salas-Pilco, S. Z. (2024). Artificial Intelligence for Enhancing Special Education for K-12: A Decade of Trends, Themes, and Global Insights (2013\u0026ndash;2023). \u003cem\u003eInternational Journal of Artificial Intelligence in Education\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40593-024-00422-0\u003c/span\u003e\u003cspan address=\"10.1007/s40593-024-00422-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Yeadon, W., \u0026amp; Hardy, T. (2024). The impact of AI in physics education: A comprehensive review from GCSE to university levels. \u003cem\u003ePhysics Education\u003c/em\u003e, \u003cem\u003e59\u003c/em\u003e(2), 025010. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1088/1361-6552/ad1fa2\u003c/span\u003e\u003cspan address=\"10.1088/1361-6552/ad1fa2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Yue, M., Lyu, W., Mifdal, W., Suh, J., Zhang, Y., \u0026amp; Yao, Z. (2024). \u003cem\u003eMathVC: An LLM-Simulated Multi-Character Virtual Classroom for Mathematics Education\u003c/em\u003e (Version 2). arXiv. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.48550/ARXIV.2404.06711\u003c/span\u003e\u003cspan address=\"10.48550/ARXIV.2404.06711\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Yunianto, W., Lavicza, Z., Kastner-Hauler, O., \u0026amp; Houghton, T. (2024). Investigating the use of ChatGPT to solve a GeoGebra based mathematics\u0026thinsp;+\u0026thinsp;computational thinking task in a geometry topic. \u003cem\u003eJournal on Mathematics Education\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(3), 1027\u0026ndash;1052. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.22342/jme.v15i3.pp1027-1052\u003c/span\u003e\u003cspan address=\"10.22342/jme.v15i3.pp1027-1052\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYusuf, A., Pervin, N., Rom\u0026aacute;n-Gonz\u0026aacute;lez, M., \u0026amp; Noor, N. M. (2024). Generative AI in education and research: A systematic mapping review. \u003cem\u003eReview of Education\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(2), e3489. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/rev3.3489\u003c/span\u003e\u003cspan address=\"10.1002/rev3.3489\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003e*Zhang, F., Li, C., Henkel, O., Xing, W., Baral, S., Heffernan, N., \u0026amp; Li, H. (2024). Math-LLMs: AI Cyberinfrastructure with Pre-trained Transformers for Math Education. \u003cem\u003eInternational Journal of Artificial Intelligence in Education\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s40593-024-00416-y\u003c/span\u003e\u003cspan address=\"10.1007/s40593-024-00416-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e By \u0026ldquo;secondary STEM education\u0026rdquo; we mean both education in the individual STEM disciplines (e.g. mathematics education) and the combination or integration of several STEM disciplines within an educational setting. We refrain from an in-depth discussion of the relationship and integration of the various disciplines (see English, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2017\u003c/span\u003e, for an overview).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e We decided not to include \u003cem\u003etechnology\u003c/em\u003e directly as a discipline to avoid irrelevant results, which are not related to technology education but used to term \u003cem\u003etechnology\u003c/em\u003e, for example to state that ChatGPT is an emerging technology. We suppose that this restriction does not exclude relevant results.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e Retrieval augmented generation (RAG) is a possibility make a generative AI system use an external data base to retrieve information.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Artificial Intelligence, Human Flourishing, SWOT analysis, scoping literature review, secondary education","lastPublishedDoi":"10.21203/rs.3.rs-6923010/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6923010/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Education for Human Flourishing aims to empower students to develop their full potential to lead a meaningful, autonomous life to the benefit of themselves and society at large. Recent technological developments call for an evaluation of the entanglement of (education for) Human Flourishing and Artificial Intelligence. Following the PRISMA guidelines, this scoping review investigates to what extent the current research on generative AI in secondary STEM education provides a solid basis for exploring the interconnection of Artificial Intelligence and Human Flourishing in STEM education. To this end, 183 eligible publications were analyzed regarding their general characteristics, research themes as well as strengths, weaknesses, opportunities and threats (SWOTs).\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cstrong\u003eResults:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e The scoping literature review reveals a focus on cognitive aspects of STEM education despite the need to broaden human capabilities in the light of generative AI. Ethical aspects are sidelined, although the SWOT analysis shows some significance of these issues. Moreover, there is a lack of research on STEM-specific theoretical frameworks and research is concentrated in the Global North, both of which might undermine an unbiased, culturally diverse development of new solutions for generative AI in secondary STEM education. The majority of current research examines AI-generated content instead of human participants, and publications focus on the performance and development of AI tools instead of their impact and application. This might hinder a human-centered approach to AI in secondary STEM education, potentially threatening human identity and meaning and thereby Human Flourishing.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e\u003c/em\u003e\u003cem\u003e Based on the results, we show that existing literature does not yet provide a suitable foundation for Human Flourishing related to Artificial Intelligence in secondary STEM education. Our findings thus point to future research perspectives necessary to strengthen Human Flourishing in STEM education and ensure a human-centered, meaningful approach to Artificial Intelligence.\u003c/em\u003e\u003c/p\u003e","manuscriptTitle":"Generative Artificial Intelligence in Secondary STEM Education in the Light of Human Flourishing: A Scoping Literature Review","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-02 06:50:49","doi":"10.21203/rs.3.rs-6923010/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"e4328fc3-b6b9-4de0-b3bf-e438d5743cd7","owner":[],"postedDate":"July 2nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-12-08T16:11:05+00:00","versionOfRecord":{"articleIdentity":"rs-6923010","link":"https://doi.org/10.1186/s40594-025-00589-5","journal":{"identity":"international-journal-of-stem-education","isVorOnly":false,"title":"International Journal of STEM Education"},"publishedOn":"2025-12-01 15:57:38","publishedOnDateReadable":"December 1st, 2025"},"versionCreatedAt":"2025-07-02 06:50:49","video":"","vorDoi":"10.1186/s40594-025-00589-5","vorDoiUrl":"https://doi.org/10.1186/s40594-025-00589-5","workflowStages":[]},"version":"v1","identity":"rs-6923010","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6923010","identity":"rs-6923010","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}